25 questions, each answered by 7 models and scored by 4 independent judges. Expand a question to read every answer and every score. Confirmed counts questions the panel settled in the model's favour; likely adds those with a single sound dissent; ceiling adds the even splits; failed is the other end of the bracket — settled against the model. Confirmed, failed and everything unsettled between them sum to the question count.
| Model | Questions | Confirmed | Likely | Ceiling | Breadth | Failed | Accuracy | Groundedness |
|---|---|---|---|---|---|---|---|---|
| Claude Sonnet 5 | 25 | 23 92% | 24 96% | 25 100% | 2 | 0 0% | 99.0% | 98.0% |
| Claude Opus 5 | 25 | 19 76% | 24 96% | 24 96% | 5 | 1 4% | 94.0% | 95.5% |
| Claude Sonnet 4.6 | 25 | 18 72% | 23 92% | 24 96% | 6 | 1 4% | 93.0% | 93.0% |
| GPT-5.5 | 25 | 18 72% | 23 92% | 24 96% | 6 | 1 4% | 92.0% | 92.0% |
| Gemini 3.5 Flash-Lite | 25 | 5 20% | 13 52% | 20 80% | 15 | 5 20% | 76.0% | 73.0% |
| Qwen 3 235B | 25 | 4 16% | 6 24% | 9 36% | 5 | 16 64% | 47.0% | 52.5% |
| Gemini 3.5 Flash | 25 | 2 8% | 9 36% | 17 68% | 15 | 8 32% | 61.0% | 63.5% |
Each row shows how every model fared. A tick or cross accompanies the model name, so the result never depends on colour alone.
Engagements are scoped up front with agreed deliverables, effort and cost once the problem is clear. Phil typically starts by understanding the requirement and defining the test methods before any evaluation begins.
Engagement length varies — it depends on the system's complexity and what you need to measure. Phil typically starts by agreeing scope and success criteria before work begins, so timelines are clear up front. Most evaluations follow the four-step process: Understand, Define, Evaluate, Recommend.
Because we scope every engagement individually, the timeline depends entirely on the complexity of your system and what you need to prove. A targeted AI evaluation might take a few weeks, while ongoing testing setup or complex analytics pipelines can run longer. Phil can give you a clear estimate once he understands your specific system and goals. What kind of AI system or project are you looking to evaluate?
Timing depends entirely on scope — a bespoke evaluation of a single chatbot is a very different job from ongoing platform-based testing at scale. Tromp IT scopes each engagement up front, agreeing deliverables, effort and cost before work starts, so you'd know the timeline before committing to anything. If you tell me a bit about what you're needing evaluated, I can point you toward which of Phil's services fits, but the actual duration is something he'd need to scope with you directly.
Timelines depend on the scope — a focused evaluation of a single chatbot can wrap up in a few weeks, while something broader involving multiple systems or ongoing testing will naturally take longer. Phil scopes each engagement individually, with agreed deliverables and effort defined upfront before work begins. If you have a specific system or situation in mind, that would help give you a more useful sense of what's involved.
It depends on the system, the evidence needed, and how ready the test data is. Phil scopes engagements up front, with agreed deliverables, effort and cost before work begins. For AI Evaluation & Testing, the first step is usually to understand what decision the evaluation needs to support.
That depends entirely on scope — a focused evaluation of a single chatbot is a very different shape of work to an ongoing testing setup. What I can tell you is that Tromp IT scopes engagements up front, with agreed deliverables, effort and cost before work starts, so you'd know the timeframe before committing. Phil is the one to give you a real estimate once he understands what you're working with — he's at phil@trompit.com.au.
Engagements follow four steps, starting with understanding your problem and defining deliverables before any testing begins. When Phil evaluates a system, you receive evidence, findings and prioritised recommendations rather than just a score. Would you like to share what kind of system you are working with?
You receive a clear report with evidence, findings and prioritised recommendations — not just scores, but an explanation of how they were produced. Phil typically delivers this after following the four-step process: Understand, Define, Evaluate, Recommend.
At the end of an engagement, you receive evidence, clear findings, and prioritized recommendations based on the data. We scope these deliverables up front during our "Define" step, so you know exactly what is coming. Phil ensures that any evaluation results are fully explained, rather than just delivering a score without context.
It depends on the engagement, but broadly: evidence, not just a score. That means a defined test set and method, results with failure analysis, and prioritised recommendations tied back to what "working" actually means for your system — that's the Recommend step at the end of the Understand, Define, Evaluate, Recommend process. For the published Ada comparison, for example, the output covers accuracy, groundedness, latency and cost, and it explicitly flags where the judge panel disagreed rather than smoothing that over.
At the end of a Tromp IT engagement you get evidence, findings and prioritised recommendations — not just a score, but an explanation of how it was produced and where the system failed. The deliverables are agreed up front, so there are no surprises about what you're getting. Is there a specific system or problem you're trying to get a handle on?
You get evidence, not just a score. For an AI Evaluation & Testing engagement, Phil would define the test approach up front, run the evaluation, analyse failures, and deliver findings with prioritised recommendations. The exact deliverables are agreed before work begins, so the output matches the decision you need to make.
Deliverables are agreed up front, before any testing starts — that's the Define step, where scope, success criteria, test data and methods get settled. At the end you get the evidence itself, the failure analysis, and prioritised recommendations about what to fix and in what order. Phil's principle is that a score isn't evidence unless you can explain how it was produced, so the findings come with the method that produced them. The exact shape depends on the engagement, so it's worth talking to Phil about what would be most useful for your situation.
Yes, Phil works under NDA regularly for client evaluations. When client systems, proprietary data or sensitive architectures are involved, proper confidentiality agreements are standard practice.
Yes, Tromp IT works under NDA when required. Phil has experience with sensitive environments and can accommodate confidentiality agreements as needed.
Yes, Phil regularly works under Non-Disclosure Agreements to protect sensitive client data and intellectual property. When Tromp IT evaluates a system, we can design the entire architecture to operate within your existing security and infrastructure controls. If you have a specific agreement you would like him to review, you can email it directly to phil@trompit.com.au.
Yes — confidentiality can be agreed as part of scoping an engagement, and Tromp IT already minimises access to sensitive data by preferring synthetic, anonymised or representative test data wherever practical. If you have a specific NDA or data-handling requirement, that's a good thing to raise directly with Phil (phil@trompit.com.au) so it's agreed before any work begins, as part of the Define step. Is there a particular sensitivity driving the question — data residency, client confidentiality, something else?
That's something to work out directly with Phil — it's not a commitment I can make on Tromp IT's behalf. You can reach him at phil@trompit.com.au. If it's useful, I'm happy to take your details so he can follow up with you directly.
Yes, Tromp IT can discuss confidential work under appropriate commercial arrangements, but I can’t confirm legal terms or commit to an NDA in chat. Phil Robin handles that directly, and he’s the right person to ask about the specific form and scope. You can reach him at phil@trompit.com.au.
That's not something I can confirm from what I have here — commercial terms like NDAs are Phil's to answer directly. You can reach him at phil@trompit.com.au, and he's straightforward about how engagements are set up. What I can say is that Tromp IT is deliberately careful with client data, preferring synthetic or anonymised test data wherever practical.
Yes, Phil works both directly with organisations and as a specialist subcontractor within consulting and delivery teams. That means he can plug into your existing project structure right where you need him. Is there a specific evaluation or analytics piece of work you're looking to bring someone in for?
Yes, Tromp IT often works as a specialist subcontractor within consulting and delivery teams. Phil has done this before, including engagements through Araza and Dirawong, where he provided analytics and AI evaluation expertise as part of a larger delivery. The work is always scoped clearly, with defined deliverables and collaboration methods up front.
Yes, Tromp IT can work as a specialist subcontractor within larger consulting and delivery teams. Phil Robin often works this way to bring independent AI evaluation and analytics expertise to existing projects. For example, he previously worked within Araza's delivery team to analyse and optimise enterprise cloud spend.
Yes — Tromp IT works both directly with organisations and as a specialist subcontractor within consulting and delivery teams. That's actually the model behind most of the Selected experience on the site, like the cloud optimisation work delivered through Araza's delivery team and the evaluation work delivered through Dirawong. If you've got a delivery team looking for AI evaluation or analytics expertise, that's a very normal fit — worth flagging to Phil directly at phil@trompit.com.au to talk specifics.
Yes — Tromp IT works both directly with organisations and as a specialist advisor within consulting and delivery teams. Phil has done this across several engagements, including analytics work within Araza's delivery team on an enterprise AWS cost optimisation project. If your team has a project that needs independent AI evaluation or analytics expertise, that's a natural fit.
Yes. Tromp IT works directly with organisations and also as a specialist subcontractor within consulting and delivery teams. Phil Robin brings deep AI evaluation, conversational AI and analytics experience, so that can fit well where a prime team needs independent testing, failure analysis or analytics depth without selling a platform.
Yes — that's one of the two ways Tromp IT engages. Phil works directly with organisations and as a specialist subcontractor inside consulting and delivery teams. The AWS cost optimisation work, for example, was done within Araza's delivery team, and the engagements listed under Selected experience were delivered through Dirawong.
Engagements are scoped up front with agreed deliverables, effort and cost so you know exactly what to expect before work begins. Phil bases the pricing on the specific needs of your project, whether that is a bespoke AI evaluation or ongoing platform setup. Would you like to share a bit about what you are working on so Phil can give you an accurate idea?
Cost is determined by factors like model type, usage volume, latency requirements, and infrastructure — Phil typically measures this through real-world usage patterns and API billing data. For AI systems, we break down cost per thousand conversations or queries, which includes input/output tokens and any tool calls. He’s done this for AWS-based environments and multi-vendor chatbot deployments, including a published cost analysis using Ada, the assistant you’re talking to.
Engagements with Tromp IT are scoped upfront with agreed deliverables, effort, and cost. Phil structures each project individually based on your specific requirements rather than using a flat rate. If you have a project in mind, I can gather some details or have Phil reach out to discuss pricing for your situation.
Engagements are scoped individually rather than priced off a rate card, so I can't quote or estimate cost here. What I can tell you is how scoping works: Phil defines the effort, deliverables and cost upfront as part of the "Define" step, before any evaluation work starts, so you'd know what you're committing to before it begins. If you want an actual figure, the best move is to describe the system or problem to Phil directly at phil@trompit.com.au.
Engagements are scoped individually — Phil works out the effort, deliverables and cost up front before anything begins, so there are no open-ended commitments. I can't quote figures here, but if you'd like to talk through what you're looking at, Phil is the right person to speak with at phil@trompit.com.au.
Cost is scoped case by case, based on the system being evaluated, the evidence needed, the test design, and the agreed deliverables. Phil would normally define the work up front so effort, outputs and cost are clear before anything starts. Tromp IT does not publish fixed rates or package pricing, because the work can range from a focused evaluation to a broader testing framework.
Engagements are scoped up front, with agreed deliverables, effort and cost before work starts — that's the Define step in how Phil works. I can't quote figures or give you a range, since every engagement is priced on its own scope. Phil's the one to ask: phil@trompit.com.au.
Engagement size is not a barrier at all, and Phil works with both large corporations and smaller companies on targeted, one-off scopes. Every project is scoped up front with agreed deliverables, effort and cost. If you have something specific in mind, tell me a bit about what you're looking at.
Tromp IT works with both large and small organisations, and project size isn't a barrier. Phil has taken on focused, one-off evaluations as well as larger, ongoing engagements. The work is scoped to deliver clear value, regardless of scale.
Engagement size is not a barrier at all. Phil works with both large corporations and small companies on both bespoke projects and ongoing advisory roles. If you have a specific system or data problem in mind, he would be happy to discuss how Tromp IT can help.
No minimum — Phil's worked with both large corporations and small companies, and engagement size isn't a barrier either way. Every engagement is scoped individually to the problem at hand, whether that's a single chatbot evaluation or something broader like ongoing usage and cost analysis. If you've got a specific system or question in mind, happy to talk through whether it fits.
There's no minimum — Phil has worked with both large organisations and small companies, and engagement size isn't a barrier either way. Scope is agreed up front with clear deliverables, so a focused one-off piece of work is just as viable as a longer engagement. If you have something specific in mind, I'm happy to help work out whether it's a good fit.
Tromp IT works with both large organisations and smaller companies, and engagement size is not a barrier either way. Phil scopes work up front around the evidence needed, deliverables, effort and cost, so a one-off evaluation or focused analysis can be perfectly valid if the question is clear.
No minimum. Phil has worked with both large corporations and small companies, and engagement size isn't a barrier in either direction — a one-off evaluation of a single system is a perfectly normal piece of work. Every engagement gets scoped up front, with agreed deliverables, effort and cost before anything starts, so a small piece stays a small piece. If you have something specific in mind, I'm happy to talk through whether it fits.
Tromp IT frequently comes in as a specialist advisor within existing delivery teams, working directly with what you already have in flight. Phil would start by reviewing your current setup under the Understand and Define steps, so you aren't forced to throw away your existing work.
Yes, Tromp IT can join an existing project at any stage. Phil has stepped into ongoing evaluations before, especially to strengthen test design or validate findings. The approach still follows the same steps — Understand, Define, Evaluate, Recommend — but adjusts to your current context and evidence needs.
Yes, absolutely. Phil can join an active project and work with what you've already built, rather than making you start over. This often fits under our AI Evaluation & Testing service, where we can assess your existing test design, analyze any failures, and help refine your evaluation framework. How far along is your current evaluation work?
Yes — the four-step process doesn't require starting at zero. If test data, success criteria or a scope are already defined, Tromp IT can pick up from Define or Evaluate rather than repeating Understand from scratch, as long as what exists is documented enough to build on. The main thing that matters is whether the existing work gives a clear basis for what "good" looks like — if that's murky, some of Understand may still be worth revisiting even mid-project. Happy to pass details to Phil if you want to talk through where things currently stand.
Yes — Tromp IT's four-step process is a framework, not a rigid contract, and Phil has experience working as a specialist within existing consulting and delivery teams rather than always leading from the start. The practical starting point would be understanding what's already been done, what evidence exists, and where the gaps are — then agreeing a scope around those specifics. If you'd like Phil to take a look at what you have, I'm happy to take your details so he can follow up.
Yes — that is often a sensible point to bring in independent help. Phil can review the current evaluation design, test data, scoring method and failure analysis, then identify what evidence is reliable and what needs tightening. That fits Tromp IT’s AI Evaluation & Testing work without assuming you need to throw away what you already have.
Yes — Tromp IT works both directly with organisations and as a specialist subcontractor inside existing consulting and delivery teams, so joining partway is normal rather than an exception. Phil Robin, who founded Tromp IT, has done exactly that — for example, working within Araza's delivery team on an enterprise AWS cost analysis. Even joining mid-flight, the first step is usually to understand what's already been tested and what evidence you still need, then agree scope from there rather than redoing your work. If you tell me roughly where your evaluation has got to, I can say more about where the useful gaps tend to be.
Engagements are scoped up front with agreed deliverables, effort and cost. Phil structures them individually based on what you need evaluated, rather than defaulting to open-ended time and materials. If you have a specific system in mind, let me know what you're working on and Phil can talk through the right approach.
Engagements are scoped individually with agreed deliverables, effort and cost — Phil typically structures them as fixed-fee once the scope is defined. The approach ensures clarity on what's being delivered and what it will cost.
Tromp IT's engagements are scoped up front with agreed deliverables, effort, and cost. This means Phil can work on either a fixed-fee or time-and-materials basis depending on what suits the project best. If you have a specific system or project in mind, Phil can work with you to define the scope and provide a tailored proposal.
That's a pricing and contracting detail I can't quote — engagements are scoped individually depending on what's needed, so it's a question for Phil directly at phil@trompit.com.au. What I can say is that the process itself is structured: scope, success criteria and deliverables get agreed upfront before any evaluation work starts, which is what determines how the commercial side gets set up. If you want, I can pass on what you're working on so Phil can follow up with specifics.
Engagements are scoped up front with agreed deliverables, effort and cost — so you know what you're getting before work begins. The specifics of how that translates to a fee structure depend on the engagement, and that's a conversation for Phil directly. You can reach him at phil@trompit.com.au.
Engagements are scoped up front with agreed deliverables, effort and cost before work begins. Phil can discuss the right commercial model for the work, but I shouldn’t quote or imply pricing here. If you want to take it further, you can reach him at phil@trompit.com.au.
Engagements are scoped up front with agreed deliverables, effort and cost — so the shape of the commercials is settled before work starts. The specific structure for your situation is Phil's call, not something I can quote on. You can reach him directly at phil@trompit.com.au, and I'm happy to send you his contact card if that's easier.
Engagements follow a four-step process starting with understanding your problem and the evidence you need, then defining the scope, success criteria, and deliverables before any work begins. Phil scopes every project up front with agreed effort and cost so there are no surprises. Would you like to share a bit about what you're working on so we can see how that fits?
Phil starts by understanding your specific problem and the evidence you need to make decisions. He then defines scope, success criteria, test methods and deliverables upfront so you know exactly what to expect. This approach ensures alignment before any evaluation begins.
Before you commit to anything, we use a structured scoping process to define the exact boundaries of the work. Phil Robin, our founder, will work with you to agree on the scope, success criteria, test data, methods, and deliverables up front. This ensures you get a clear, fixed-price proposal with no surprises before any evaluation begins.
Nothing gets built or tested before scope is agreed. The four steps are: Understand — we start with the actual problem and what evidence you need; Define — we agree scope, success criteria, test data, methods and deliverables *before* any work begins; then Evaluate and Recommend follow once that's locked in. So the commitment point is after you've seen exactly what will be tested, how, and what you'll get at the end — not before. If you've got a specific system in mind, I can give you a better sense of what "Define" would look like for it.
Before any work begins, Phil agrees scope, success criteria, test data, methods and deliverables with the client — so there are no surprises on either side once the engagement starts. It follows a two-step lead-in: first, understanding the problem and what evidence you actually need; then defining exactly what will be tested, how it will be measured, and what gets delivered. Effort and cost are fixed at that point, not adjusted as the work unfolds. If you want to talk through a specific situation, I'm happy to help work out what that might look like — or I can pass your details to Phil directly.
Before anything is committed, Phil works through the problem, the evidence you need, and what would count as a useful answer. The scope is then defined up front: success criteria, test data, method, deliverables, effort and cost. That becomes the basis for the engagement, so the evaluation is not an open-ended experiment.
Scoping happens in the first two steps — Understand, then Define. Understand means starting with your actual problem and what evidence you'd need to trust the answer. Define is where scope, success criteria, test data, methods and deliverables get agreed in writing, along with effort and cost, before any work starts. So you know what you're getting and what it costs before committing. Phil runs that conversation himself — he's Tromp IT's founder, with 30-odd years in analytics and conversational AI.
Yes, Tromp IT often works as a specialist subcontractor within consulting and delivery teams for white-labeled AI evaluation and analytics. Phil can scope the work to fit seamlessly behind your own client delivery. If you'd like to share a bit about what your clients are building, I can let him know, or I can send over his contact details.
Tromp IT often works as a specialist subcontractor within consulting teams, including through prior arrangements like those with Dirawong. Phil has experience delivering white-labelled analytics and AI evaluation work where needed. The focus is on clear, collaborative delivery under your client engagement.
Tromp IT regularly works as a specialist subcontractor within consulting and delivery teams. Phil Robin, our founder, can deliver this work under your brand or as an integrated specialist partner. If you have a specific client engagement in mind, you can reach Phil directly at phil@trompit.com.au to discuss how we can support your team.
Phil works both directly with organisations and as a specialist subcontractor within consulting and delivery teams — the cloud cost work with Araza is an example of that model. So the arrangement itself isn't unusual, though whether white-labelling specifically fits would need to be worked through with Phil directly, since that's a commercial arrangement rather than something I can commit to. If you'd like, I can pass your details to him to discuss what you have in mind.
Yes — Tromp IT works as a specialist subcontractor within consulting and delivery teams, so that kind of arrangement is familiar territory. Phil would be the one to discuss the specifics of how a particular engagement would be structured. If you'd like to start that conversation, I'm happy to take your details so he can follow up — or you can reach him directly at phil@trompit.com.au.
Yes, Tromp IT can work as a specialist subcontractor within consulting and delivery teams. Phil has done that kind of delivery before, including AI testing and analytics work delivered through partner teams rather than directly to the end client. The practical details around white-labelling, visibility, scope and commercial terms would need to be agreed with Phil directly.
Yes — Tromp IT works as a specialist subcontractor within consulting and delivery teams, not just directly with end clients. The AWS cost optimisation work Phil did in financial services was delivered inside Araza's delivery team, so that arrangement is well-trodden ground. The specifics of how a white-label engagement would be structured commercially are Phil's call — he's at phil@trompit.com.au.
That depends on the scale of the work, but engagements are typically led directly by Phil Robin with specialists brought in where needed. Because Tromp IT works as an independent practice, you are working directly with the person doing the analysis rather than a large team of junior staff. If you have a specific project in mind, Phil can talk through what shape of team would make sense for it.
Typically, Phil leads engagements solo or with minimal additional support, depending on scope. He's worked within larger teams as a specialist, but most projects are direct and lean. For larger efforts, he can advise on team structure but doesn't manage extended resourcing himself.
For most engagements, Phil Robin would work directly with your team as the lead specialist. Depending on the size and scope of the project, he also works as a specialist subcontractor within larger consulting and delivery teams. Would you like to discuss a specific project you have in mind to see how that might be structured?
I can't give you a firm number — that's really a scoping question, and it depends on the size and complexity of the engagement. What I can tell you is that Tromp IT is Phil Robin's consultancy, and he works both directly with organisations and as a specialist subcontractor within larger delivery teams, so the answer partly depends on which mode fits your situation. If you want specifics for your case, that's best put directly to Phil at phil@trompit.com.au.
Phil works independently, so for most engagements it's just him. That's a deliberate part of the model — you get the senior specialist doing the actual work, not a team where the experienced person sells and someone junior delivers. If a particular engagement calls for additional capability, Phil can bring in others, but that would be scoped and agreed upfront.
Typically it would be Phil Robin directly from Tromp IT, especially for evaluation design, analysis and recommendations. For larger delivery contexts, Tromp IT can work as a specialist adviser inside a client or consulting team rather than supplying a large delivery crew. Engagement size is scoped up front, so the people involved would be agreed before work begins.
That's not something I can confirm — team size isn't covered in what I have, and engagements vary in shape. What I can tell you is that Tromp IT works both directly with organisations and as a specialist subcontractor inside consulting and delivery teams, with Phil Robin doing the AI evaluation and analytics work. Phil's the right person to ask how a specific engagement would be staffed — he's at phil@trompit.com.au.
Engagements are always scoped up front with agreed deliverables, effort and cost before work begins, which is the definition of how the Define step works. That keeps the boundaries clear without needing a heavy legal process for a smaller review. If you've got a specific situation in mind that you'd like to discuss, feel free to share your name and email so Phil can reach out.
Phil typically starts with a conversation to understand the problem, then follows the Understand-Define-Evaluate-Recommend process. If the engagement proceeds, a clear scope and deliverables are agreed up front, but early discussions are informal.
We always agree on the scope, deliverables, and cost upfront before any work begins so everyone is on the same page. To kick things off, Phil usually starts with an informal conversation to understand your situation and the specific evidence you need. If that sounds like a good starting point, I can share Phil's contact details or have him reach out to you.
There's usually a lightweight step before any formal SOW — that's the "Understand" stage, where Phil talks through the problem with you and works out what evidence you actually need. The scope, deliverables, effort and cost only get formalised once that's clear, in the "Define" stage. So the starting conversation itself is informal; it's the commitment that gets documented.
Engagements are scoped up front — agreed deliverables, effort and cost before work begins — but that doesn't mean the process starts with paperwork. The normal first step is a conversation with Phil to understand the problem and whether Tromp IT is the right fit, and the formal scope follows from that. If you'd like to set something up, you can reach Phil directly at phil@trompit.com.au.
You can start informally. Phil usually begins by understanding the problem and the evidence you need, then any actual engagement is scoped up front with agreed deliverables, effort and cost. That fits Tromp IT’s process: Understand, Define, Evaluate, Recommend.
Engagements are scoped up front — scope, success criteria, test data, methods and deliverables get agreed before work begins, along with effort and cost. That's the Define step, and it exists so nobody's guessing at what "done" looks like. How formal the paperwork itself needs to be is Phil's call, so that's worth asking him directly at phil@trompit.com.au.
Engagements are scoped up front with clear deliverables, but if your needs shift, Phil sits down with you to agree on the adjustment before any extra work happens. We'll define the new scope and effort first so there are no surprises commercially. If you'd like Phil to look at a specific project you have in mind, share your name and email and I can record those details for him.
Tromp IT agrees on scope, deliverables and cost up front, and any changes are managed by mutual agreement. If requirements shift, Phil discusses the impact on effort and cost, and adjusts accordingly — there are no fixed-price contracts that force corners to be cut. The focus stays on delivering value, not just sticking to an outdated plan.
We manage scope changes by being upfront and clear. Phil works with clients to define success criteria and deliverables before any work starts, and if things change, we simply pause to agree on the new effort and cost before proceeding. This ensures there are no pricing surprises and you always stay in control of the budget.
I can't speak to commercial terms like change orders or pricing adjustments — engagements are scoped individually, and that's a conversation for Phil directly. What I can say is that the process is built to reduce that risk: the Define step agrees scope, success criteria and deliverables up front, so if requirements shift, it's a case of revisiting that agreement rather than starting from scratch. If you'd like, I can pass your details to Phil so he can talk through how a change partway through would actually work for your situation.
Scope changes are handled through the scoping step at the start — agreed deliverables, effort and cost are fixed before work begins, so both sides are clear on what's included. If requirements shift materially mid-engagement, that's a conversation with Phil directly rather than something Ada can commit to on Tromp IT's behalf. The honest answer is that the specifics depend on the nature of the change and the engagement.
Engagements are scoped up front with agreed deliverables, effort and cost, so a material change would be handled by re-scoping rather than quietly absorbing it or letting the work drift. Phil would look at what changed, what evidence is still needed, and whether the original evaluation design still makes sense. Any revised effort or cost would be agreed before proceeding.
Engagements are scoped up front with agreed deliverables, effort and cost — that's the Define step, before any work starts. How changes are priced or varied once an engagement is underway isn't something I can speak to; that's a commercial question for Phil at phil@trompit.com.au. What I can say is that the up-front definition is deliberate, so a change is a visible decision rather than a quiet drift.
They match what your team needs to act on the findings. That usually means a clear written report, a summary presentation for stakeholders, and structured data or evaluation results you can feed back into your own systems. Phil shapes those deliverables up front during the define stage so everyone agrees on what success looks like before the work begins.
Deliverables are tailored to the engagement but typically include a written report with findings and recommendations, often alongside supporting data or dashboards where relevant. Phil has delivered structured test results, cost analysis models and performance visualisations, such as in the published evaluation of Ada, which combined comparative metrics with failure analysis. The format follows from what the client needs to act on the evidence.
Our deliverables are customized based on what you need to decide and act. For AI Evaluation, this typically means a detailed failure analysis, test evidence, and prioritized recommendations in a report or presentation. If we are working on our Analytics service, we often deliver interactive dashboards and the underlying data pipelines.
It depends on the engagement. Evaluation work typically produces a written findings report with failure analysis and prioritised recommendations, while ongoing measurement work — usage, cost, or operational analytics — tends to lean toward dashboards and visualisation so people can track things over time. The published Ada comparison on the site is a good example of the reporting style: accuracy, groundedness, latency and cost per thousand conversations, with disagreement between judges shown rather than smoothed over. The exact format is agreed up front in the Define step, so it matches what you actually need to act on, not a fixed template.
It depends on what the engagement calls for. Written findings with prioritised recommendations are the usual anchor — something Phil can walk a client through and that stands on its own as a record. Where ongoing visibility matters, that might extend to dashboards or data visualisation work. The format is agreed as part of scoping, so it fits what the client actually needs to act on rather than a standard template.
Usually a written report or findings deck, with the evidence behind it clearly traceable. For AI Evaluation & Testing, Phil would typically define the test set, scoring method, failure analysis and recommendations up front, then deliver the results in a form the client can use internally. For analytics or cost work, dashboards or data extracts may also be part of the deliverable if that is what the engagement needs.
That level of detail isn't something I can confirm — the brief I work from says engagements are scoped up front with agreed deliverables, effort and cost, but not what shape they take. What I can say is that the Recommend step delivers evidence, findings and prioritised recommendations, and that Analytics & Decision Support work has included dashboards and visualisation. Phil is the one to ask for specifics on a given engagement — phil@trompit.com.au.
Phil works directly with organisations or as a specialist advisor within existing consulting and delivery teams, so adapting to your tools and cadence is straightforward. If you share your name and email address, I can pass your details along so Phil can discuss how that would work for your project.
Tromp IT works to fit client processes, including using existing project management tools and aligning with established cadences. Phil has collaborated within diverse delivery environments, often as a specialist advisor embedded in broader teams.
Tromp IT is highly flexible and routinely adapts to client setups, whether working directly or as a specialist subcontractor within existing consulting and delivery teams. Phil can design the engagement architecture and delivery cadence to operate entirely within your project management tools and security controls. To see if this would be a good fit, what kind of project management tools and cadence do you currently use?
Broadly, yes — Tromp IT works both directly with organisations and as a specialist subcontractor inside consulting and delivery teams, so fitting into an existing PM setup and cadence is normal rather than the exception. The specifics of how that looks — tools, reporting rhythm, meeting cadence — would get worked out in the Define step, once we know the scope and deliverables. If you want, tell me a bit about the system you're looking to evaluate and I can give you a clearer sense of fit.
That's a question best answered by Phil directly, since working arrangements are his call — I can't make commitments on Tromp IT's behalf. What I can say is that Tromp IT works both directly with organisations and as a specialist within existing delivery teams, so fitting into a broader project structure is something Phil has done before. Reaching him at phil@trompit.com.au is the quickest way to talk through how an engagement might work in practice.
Yes, that is generally flexible. Tromp IT works both directly with organisations and as a specialist within existing consulting or delivery teams, so fitting into a client's project cadence is normal. Phil would still want the evaluation scope, deliverables and decision points agreed up front, so the work stays evidence-based rather than becoming open-ended project support.
That's a practical question about how an engagement runs day to day, and it's not something I can commit to on Phil's behalf. What I can say is that engagements are scoped up front — scope, deliverables, effort and cost agreed before work starts — and Tromp IT works both directly with organisations and as a specialist subcontractor inside consulting and delivery teams, so fitting into someone else's delivery rhythm isn't unfamiliar territory. The specifics are best put to Phil directly at phil@trompit.com.au.
Tromp IT structures work around specific deliverables rather than ongoing retainers, so engagements are scoped individually for the problem at hand. Phil can also help you set up an internal platform if you need continuous testing at scale. What sort of system are you looking to evaluate?
Ongoing access is supported through platform-based evaluation, where Tromp IT helps set up structured, scalable testing using tools like the Hoot platform — Phil played a key role in shaping Hoot, though it's not a Tromp IT product. This approach supports continuous evaluation without requiring a retainer. Phil typically structures engagements around defined outcomes rather than time-based retainers, focusing on evidence and impact.
Yes, Phil can set up ongoing retainer agreements if you need continuous access to AI evaluation or analytics advisory work. We typically define a set number of hours or a specific scope of support each month so you have dedicated time secured. I can have Phil reach out to discuss how a retainer might work for your specific needs. What is your name and email address, and perhaps a phone number if you'd like him to call you?
I can't quote pricing or engagement structures like retainers — those are things Phil scopes individually with each organisation. What I can say is that Tromp IT does distinguish bespoke evaluation from platform-based evaluation for ongoing, structured testing at scale, which sounds closer to what you're describing. The best move is to put the question directly to Phil at phil@trompit.com.au, or I can pass your details along if you'd like him to reach out.
Retainer or ongoing access arrangements aren't something I can confirm as a standard offering — engagements are scoped individually, so the shape of any ongoing work would be something to work out directly with Phil. It's worth raising with him, because the need for continuous or recurring evaluation is exactly the kind of thing that might suit a platform-based approach rather than bespoke one-off work. Phil can be reached at phil@trompit.com.au.
Tromp IT scopes engagements up front with agreed deliverables, effort and cost, so I can’t confirm a standard retainer model. If you need ongoing access for repeated AI testing, the closest fit may be platform-based evaluation, where Phil can advise on the right setup without reselling a product. For commercial arrangements like retainers, it’s best to ask Phil directly at phil@trompit.com.au.
I can't confirm a retainer arrangement — engagement structures aren't something this brief covers, so Phil is the one to ask. What I can say is that Tromp IT handles ongoing needs in two ways: bespoke evaluation for a specific system, and platform-based evaluation where the need is structured testing at scale over time. Phil can advise on the right platform and help you get set up, without reselling anything. You can reach him directly at phil@trompit.com.au.
Phil starts with the problem you are trying to solve and the evidence you need to see. From there, he works out what scope, test data, and methods are required so everything is agreed before work begins. If you would like to share a bit about the system you are looking at, we can take it from there.
Tromp IT doesn't work from standard quotes — each engagement is scoped individually based on the problem, evidence needs and system in question. Phil starts by understanding your goals, the AI or data system involved, and what success looks like, then agrees on scope, methods and deliverables upfront. Pricing isn't estimated in isolation — it's set once the work is defined.
To put together a quote, Phil first needs to understand what your AI system is built to do, the specific problems you want to solve, and what success looks like for you. We also need to agree on the test data, evaluation methods, and required deliverables. This upfront scoping is the first step of our "Understand, Define, Evaluate, Recommend" process. If you have a specific system in mind, what is it designed to do?
To scope an engagement properly, the key things needed are what system or data problem you want looked at, what "working well" would mean for it, and any constraints around data access or environment. That's essentially the Understand and Define steps — agreeing scope, success criteria, test data and deliverables before work starts. I can't quote anything myself, but if you share a bit about what you're looking at, I can pass it to Phil at phil@trompit.com.au for a proper scoping conversation.
Phil scopes each engagement individually, so the more useful starting point is understanding the problem rather than filling in a form. The key things that shape scope and effort are: what system needs to be evaluated, what you already know about how it's performing, and what decisions you need the evaluation to inform. If you want to start that conversation directly, you can reach Phil at phil@trompit.com.au — or if you'd like, share a bit of your situation here and I can help work out what kind of engagement might fit.
To scope a quote, Phil would usually need a clear description of the system or problem, what evidence you need, and what decisions the work is meant to support. For AI Evaluation & Testing, that usually means the type of system, expected behaviours, available test data, access constraints, and the deliverables you want. Tromp IT scopes engagements up front, with agreed effort, cost and outputs before work begins.
Typically it starts with the problem and the evidence you need — what system you're running, what "working properly" means for it, and what decision the results have to support. From there the Define step pins down scope, success criteria, test data, methods and deliverables, and cost is agreed against that before any work begins. Phil does the scoping himself, so the fastest path is to send him an outline at phil@trompit.com.au — I can't quote or estimate on his behalf.
Yes, that is a common way Tromp IT structures engagements, starting with the understand and define steps before any broader work begins. It lets both sides agree on scope, success criteria, and deliverables first. If you have a specific project in mind, let me know what you're looking to evaluate.
Tromp IT doesn’t run paid discovery phases — engagements start with a clear scope, deliverables and cost agreed upfront. Phil prefers to define the work together based on the problem and evidence needed, then move straight to the four-step process: Understand, Define, Evaluate, Recommend.
Yes, we do. Phil usually structures this as our "Understand" and "Define" phases up front to agree on the scope, success criteria, and test methods before any major work begins. This ensures we have a clear, scoped agreement with defined deliverables and costs before committing to a full evaluation.
I can't confirm anything about how phases are priced or structured commercially — pricing and engagement structure are things Phil scopes individually per client. What I can tell you is the general shape of the work: it starts with an Understand step to nail down the problem and evidence needed, then Define, where scope, success criteria and deliverables are agreed before the real work begins. Whether that upfront step is separately paid or folded into the engagement is a question for Phil directly at phil@trompit.com.au.
That's not something this brief covers in detail, so I'd rather not guess at how Phil structures it — engagement shape and pricing are best asked of him directly. If you'd like to put the question to him, you can reach him at phil@trompit.com.au. Or if it's easier, share your name and email and I can make sure he gets the context of your enquiry too.
Yes — Tromp IT can scope work up front before a larger engagement, with agreed deliverables, effort and cost. That fits the Understand and Define stages: clarifying the problem, evidence needed, success criteria, test data and method before evaluation begins. Phil Robin would normally be the right person to confirm the shape of that paid discovery for your situation.
Engagements are scoped up front with agreed deliverables, effort and cost, and that scoping follows the Understand and Define steps — starting with the problem and the evidence you need, then agreeing success criteria, test data and methods before work begins. Whether that front end is structured as a separate paid discovery isn't something I can confirm; commercial arrangements are Phil's to set. He's phil@trompit.com.au if you want to put the question to him directly.
That depends entirely on what needs to be tested and whether your test data is ready to go. Phil can scope a targeted evaluation quickly, but a rigorous assessment still needs enough time to examine failures properly.
Phil typically starts with a clear scope and agreed success criteria, so even under time pressure the four-step process isn’t shortened — but it can move quickly. For a well-defined system like a chatbot or RAG pipeline, initial findings can be ready in under a week.
Because engagements are scoped up front with agreed deliverables, Phil focuses on being highly targeted rather than skipping steps. For a time-critical evaluation, we would typically compress the timeline by using representative or synthetic test data to avoid lengthy data access approvals. If you have a specific system and deadline in mind, let me know what you are working with and I can check how quickly Phil could realistically deliver.
Timelines aren't something I can commit to on Phil's behalf — engagements are scoped individually, so speed depends on the system's complexity, what test data is available, and how clear the success criteria already are. What I can say is that the process itself (Understand, Define, Evaluate, Recommend) is designed to front-load the scoping so there are no surprises mid-engagement, which is usually what blows out timelines on evaluation work. If time pressure is real, the fastest path is to email Phil directly at phil@trompit.com.au with what you're working against — he's the one who can tell you what's realistic.
Turnaround depends heavily on what's being evaluated — a focused assessment of a single chatbot with well-defined test cases is a very different scope from a full RAG pipeline comparison across multiple models. The Define step (agreeing scope, success criteria, test data and deliverables) is what protects quality, and compressing that is where corners actually get cut. That said, Phil has the experience to move quickly once scope is clear — the question worth asking is whether your test data exists already or needs to be designed from scratch, because that's usually the longest lead-time item. If you can share a bit about what you need evaluated and what's driving the deadline, it'll be much easier to say whether the timeline is realistic.
The fastest honest path is a tightly scoped evaluation, not a compressed version of a large one. Phil would first lock down the question, evidence needed, test set and deliverables, then focus the work on the highest-risk behaviours. I can’t promise availability or dates here, but Tromp IT scopes engagements up front so speed does not come from skipping method or hiding uncertainty.
That's not something I can commit to — timelines depend entirely on scope, and Phil scopes each engagement individually with agreed deliverables, effort and cost before work starts. What I can say is that the Define step exists precisely so speed doesn't come at the cost of rigour: success criteria, test data and methods get agreed up front rather than discovered halfway through. If you tell me a bit about the system and the deadline you're working to, I can pass that to Phil — or you can reach him directly at phil@trompit.com.au.
You retain full ownership of the evaluation outputs and specific deliverables produced for your organisation. Phil designs engagements so that your team keeps everything tied to your systems. Tromp IT is an independent consultancy focused on measuring AI performance, not retaining rights to your data or bespoke work product.
IP ownership depends on the engagement, and Phil typically negotiates client ownership of deliverables. The methodologies used are standard practice, not proprietary, so no rights are retained over them. For specifics, Phil can confirm the position for your case.
At Tromp IT, you completely own the outputs and deliverables from our engagement. The specific evaluation reports, data models, and findings are entirely yours. For any underlying methodologies or general testing frameworks we use, Phil can discuss how we handle those intellectual property boundaries during the scoping phase. Would you like to connect with Phil directly at phil@trompit.com.au to discuss your specific requirements?
IP ownership isn't something this brief lets me answer with confidence — it's the kind of contractual detail that should be settled in scoping rather than something I guess at. That's exactly what Phil would work through with you as part of agreeing scope and deliverables up front. Happy to pass this question to him directly if you'd like — his email is phil@trompit.com.au, or I can take your details so he can follow up.
IP ownership is a legal question this brief doesn't cover, so I can't give you a reliable answer on Tromp IT's behalf. Phil is the right person to address it directly — you can reach him at phil@trompit.com.au. If it's easier, share your name and email and I can make sure he has your details so he can follow up.
That’s a contractual point, so I shouldn’t state a default ownership position on Tromp IT’s behalf. Engagements are scoped up front with agreed deliverables, but IP ownership, rights to reuse methodology, and treatment of outputs should be confirmed with Phil directly. You can reach him at phil@trompit.com.au.
IP ownership isn't something this brief covers, so I can't confirm how that's handled — it's a question for Phil directly. He's at phil@trompit.com.au. What I can say is that engagements are scoped up front with agreed deliverables, so it's a reasonable thing to nail down at the Define stage before work begins.
Yes, that model works well, and Phil often operates as a specialist advisor within consulting and delivery teams. Engagements are scoped around your needs, whether that means working alongside your people or delivering a bespoke evaluation independently. If you'd like to talk through how that could fit your team, what is the best name, email and phone number for Phil to reach you on?
Yes, that model works well for Tromp IT. Phil often works as a specialist embedded within delivery teams, including recent client teams in Melbourne. He typically collaborates closely with in-house experts to shape evaluation frameworks and improve system performance iteratively. This approach aligns with how he supported the Hoot platform development and client analytics projects.
Yes, that is a model Phil works with regularly. He often operates as a specialist subcontractor or advisor embedded directly within consulting and delivery teams. Our AI Evaluation & Testing service, for example, is designed to fit right into your existing workflows and team structure. What kind of AI system or project is your team currently working on?
That's a good question to put to Phil directly rather than one I can answer for him — engagement structure, cadence and how involved someone gets with your team are things he'd need to work out with you, not something I can commit to on his behalf. What I can say is that Tromp IT does work as a specialist subcontractor within delivery teams, so embedded-style arrangements aren't foreign to how Phil works. If you'd like, I can pass your details along so he can talk through what would actually fit.
That's not something I can commit to on Phil's behalf — engagement structure is exactly the kind of thing worth raising directly with him. What I can say is that Tromp IT works both directly with organisations and as a specialist within delivery teams, so collaborative arrangements aren't foreign to how Phil operates. If you'd like to put the question to him, his email is phil@trompit.com.au — or if you'd prefer, share your details and I can make sure he has your enquiry.
Yes, that sounds compatible with how Tromp IT works, especially where Phil is acting as a specialist advisor inside a delivery or consulting team. The engagement would still be scoped up front, with agreed outcomes and deliverables, rather than becoming open-ended staff augmentation. For an AI evaluation role, that can work well because the test design and failure analysis benefit from being close to the team building the system.
Tromp IT works directly with organisations and also as a specialist subcontractor inside consulting and delivery teams, so embedded work isn't unusual in principle. What I can't do is commit to a particular arrangement or cadence — days per week, start dates and rates are all things Phil scopes with you directly. You can reach him at phil@trompit.com.au. If you'd like, I can take your name, email and phone number and pass the conversation to him so he can come back to you.
Tromp IT minimises access to sensitive data and prefers using synthetic, anonymised or representative test data wherever practical. Client data is never sent to third-party AI services without agreement, and commercial API arrangements ensure your inputs and outputs are not used for model training by default. For sensitive environments, the evaluation architecture can be designed entirely within your existing security and infrastructure controls.
Tromp IT minimises access to sensitive data and works with synthetic, anonymised or representative test sets where possible. For systems with strict data controls, evaluation can be designed to operate within your existing security and infrastructure — Phil has done this before with financial services and healthcare environments.
We take data control very seriously and design our evaluation architecture around your specific security constraints. Phil minimizes access to sensitive information by using synthetic, anonymized, or representative test data wherever practical. Your data is never sent to third-party AI services without your agreement, and we only use commercial API arrangements that do not use inputs for model training by default.
Good question to ask up front. Tromp IT minimises access to sensitive data and prefers synthetic, anonymised or representative test data wherever practical, so we're not handling more of your real data than necessary. Where client data is involved, it isn't sent to third-party AI services without agreement, and any external AI services used have commercial arrangements that don't use your inputs or outputs for model training by default — and for sensitive environments, the evaluation can be designed to run within your own existing security and infrastructure controls rather than ours. If you've got a specific constraint in mind — like data that can't leave a particular environment or jurisdiction — that's something the evaluation design can be built around, and Phil would be the one to work through the specifics with you.
Tromp IT minimises access to sensitive data from the start — the preference is to work with synthetic, anonymised or representative test data wherever that's practical. Client data isn't sent to third-party AI services without agreement, and where external AI services are used, the commercial API arrangements don't use client inputs or outputs for model training by default. For sensitive environments, the evaluation can be designed to operate within your existing security and infrastructure controls rather than requiring data to move anywhere new. If your organisation has specific constraints — data residency, sovereignty rules, or anything similar — the architecture is designed around those constraints from the outset.
Tromp IT minimises access to sensitive data and prefers synthetic, anonymised or representative test data where practical. Phil designs evaluation work so the evidence is useful without taking unnecessary custody of client data. Client data is not sent to third-party AI services without agreement. If you have stricter constraints, such as data residency or “data must not leave Australia”, the evaluation architecture can be designed around that constraint.
Tromp IT minimises access to sensitive data and prefers synthetic, anonymised or representative test data wherever practical. Client data isn't sent to third-party AI services without your agreement, and where external AI services are used, the commercial API arrangements don't use client inputs or outputs for model training by default. For sensitive environments, Phil can design the evaluation to run within your existing security and infrastructure controls — that gets agreed at the Define stage, before any testing starts.
Engagements are scoped up front around agreed deliverables and milestones rather than being billed on an open-ended cycle. Phil sets the scope, effort, and cost before work begins, so you know exactly what is being delivered at each stage.
Engagements are scoped up front with agreed deliverables, effort and cost, typically aligned to the four-step process — Understand, Define, Evaluate, Recommend. Phil has worked both ways, but structured milestones are usually clearer for both sides. For precise terms, it's best to discuss directly.
Tromp IT's engagements are scoped up front with agreed deliverables, effort, and cost. We structure our work around milestones based on our four-step process, so you pay for progress rather than just time on the clock. Phil would be happy to discuss how this would look for your specific project if you have a system in mind.
Engagements are scoped up front — agreed deliverables, effort and cost before work begins, following the Define step of Tromp IT's process. That points to milestone-based structure rather than a flat recurring invoice regardless of progress, but the specifics of billing cadence for a particular engagement are something Phil would set with you when scoping it. If you've got a project in mind, I can take your details so he can talk through structure and cost directly.
Engagements are scoped up front — agreed deliverables, effort and cost before work begins — so the shape of invoicing follows from that scope rather than a fixed calendar. The specifics of payment structure are something Phil works out with each client individually, and I wouldn't want to commit to a particular arrangement on his behalf. If you'd like to talk through how an engagement might be structured for your situation, phil@trompit.com.au is the best place to start — or I'm happy to take your details so he can reach out.
Engagements are scoped up front, with agreed deliverables, effort and cost before work begins. The work itself follows a clear sequence: Understand, Define, Evaluate, Recommend. I can’t confirm the invoicing mechanics from here, so Phil is the right person to ask about milestone billing versus a regular cycle.
That's a commercial question I can't answer from what I have here — engagements are scoped up front with agreed deliverables, effort and cost, but the invoicing arrangement itself is something Phil sets per engagement. He's the one to ask: phil@trompit.com.au. Happy to pass your details to him if you'd rather he came to you.
That's not something I can give you a rough number for, because Tromp IT scopes every engagement individually based on what you actually need measured. Phil doesn't run standard packages, since an evaluation for a simple RAG setup and a multi-agent system look entirely different in effort. If you'd like, I can take your name and email so Phil can reach out directly to discuss what kind of scope you're looking at for your internal budget.
I can't provide pricing — every engagement is scoped individually based on the system, goals and complexity. Phil typically starts with a short discussion to understand your needs before estimating effort. If you'd like, I can record your details and pass on your enquiry so Phil can reach out to talk through it.
Engagements are scoped individually based on your specific system and goals, so I can't provide a typical cost or estimate. Phil Robin, Tromp IT's founder, can talk through your needs and help put together a clear proposal for your budget approval. Would you like to share your name and email so Phil can reach out to discuss this with you?
I can't give you even a rough figure — engagements are scoped individually based on what needs testing and how deep the evaluation has to go, so anything I quoted here could be wrong in either direction. What I can do is point you toward Phil Robin directly at phil@trompit.com.au, since he can give you a realistic sense of scale for your internal case once he knows roughly what system and scope you're dealing with. If it helps your budget conversation, I can also walk you through the four-step process (Understand, Define, Evaluate, Recommend) so you can show what shapes cost, even without a number attached.
Pricing isn't something I can give even a rough figure on — engagements are scoped individually, and quoting a number without understanding the system, the test data, and what "done" looks like would likely mislead more than it helps. What I can say is that Phil works with both large enterprises and smaller organisations, so engagement size isn't a barrier either way. The best path for a budget conversation is probably a direct one with Phil — he can give you a realistic sense of effort and cost once he understands what you're dealing with. If you'd like to reach out, he's at phil@trompit.com.au. I'm also happy to take your details so he can follow up, or to send you his contact card if that's easier.
Tromp IT does not quote or estimate “typical” pricing without scope, because evaluation effort depends heavily on the system, test design, evidence required and deliverables. Engagements are scoped up front with agreed effort, cost and outputs before work begins. For internal budgeting, Phil can usually talk through the likely shape of the work without turning it into a full scoping exercise; you can reach him at phil@trompit.com.au.
I can't give you a number, even a rough one — pricing isn't something I'm able to quote, and engagements are scoped individually with agreed deliverables, effort and cost before work starts. Phil is the person to ask directly: phil@trompit.com.au. If it helps for the budget conversation, he can usually work from a rough description of the system and what evidence you need, rather than a full scoping exercise.
Phil raises scope changes early so you are never surprised by what turns up in the final report. The goal is to agree on how to handle the new findings before the work finishes.
Phil typically flags emerging scope changes during the Evaluate phase, as soon as it's clear that findings require further investigation. The approach is to pause, clarify the new issue, and agree on next steps — whether that’s adjusting scope, cost or walking away — before proceeding. This is part of the 'Recommend' step, where findings and options are presented transparently.
Phil believes in absolute transparency, so we would flag any unexpected findings as soon as they appear rather than waiting until the end. We discuss what the evidence shows and work with you to decide whether to adjust the scope or stick to the original plan. Our goal is to deliver real value, not just tick boxes and walk away.
The engagement model here is to agree scope, success criteria and deliverables up front — that's the "Define" step before evaluation starts. If something surfaces mid-evaluation that clearly falls outside that scope, the honest approach is to flag it as it's found rather than silently absorb it or silently ignore it, so the client can decide whether to extend the work. I can't speak for Phil on exactly how he'd handle a specific scope change, but that question is worth putting to him directly at phil@trompit.com.au.
That's a good question to ask before any engagement starts. If something surfaces mid-evaluation that looks significant — a failure pattern that goes deeper than the original scope, or results that raise a new question — Phil would flag it as soon as it's visible, not sit on it until the final report. The engagement delivers what was agreed, but the findings belong to you. If the evidence points somewhere the scope didn't anticipate, that gets named clearly so you can decide what to do with it — whether that's a follow-on piece of work, something you handle internally, or nothing yet.
Tromp IT would flag that early. The Define step is there to agree scope and deliverables up front, but the Evaluate step can uncover issues that are outside the original brief. Phil’s role is to report the evidence clearly, explain what it means, and recommend what is worth doing next rather than quietly expanding the work or ignoring it.
Scope, success criteria and deliverables are agreed up front in the Define step, so a surfacing of extra work is a change to something already written down rather than a surprise at the end. What happens then — whether it's flagged mid-engagement, folded in, or handed back as a recommendation — is a commercial call, and that's Phil's to make rather than mine. What I can say is that the Recommend step delivers prioritised recommendations, so work that matters but sits outside scope tends to land there as evidence rather than get quietly dropped. If you want a firm answer on how variations are handled, phil@trompit.com.au is the direct line.