Frequently asked questions
Straight, honest answers to the questions buyers actually ask — no hype, and clear about what the product can and can't do today.
Where a finding or claim matters, the product is built so you can check it yourself — every finding traces to your documents, and the same documents produce the same report. We don't ask you to trust us; we show you why you can.
Is this just ChatGPT?
No. ChatGPT is a general chatbot — it generates a plausible answer that changes every time and cites nothing you can verify. Project Intelligence runs a fixed assurance method over your documents using a rule engine, not a chat model. It produces the same report twice from the same documents (with a digest to prove it), and every finding points back to a specific place in your files. You can ask ChatGPT a question; you can't put its answer in front of a steering committee.
Is this just a scraper?
No. It analyses the documents you upload — nothing is scraped from you or about you. Any public reference material the product uses is licensed and obtained by permission, never harvested.
Does it make things up (hallucinate)?
No. Findings aren't written by a generative model, so there's nothing to hallucinate. Every finding traces to your documents, and where something important is missing, the report says so explicitly rather than guessing. A finding that can't be traced to your documents doesn't appear.
How do I know the findings are real?
You can check every one. Each finding shows its supporting evidence and a locator pointing to where it appears in your documents. The verdict is built from a visible gate and a coverage matrix, not a hidden score, and a methodology appendix documents how the review was conducted. You're given the evidence to test the conclusion, not asked to take it on faith.
Can I see where findings come from?
Yes. Every finding is recorded with an ID, the review area it belongs to, its evidence, and a source locator. The appendix shows the full coverage matrix — every area reviewed, including those that passed — so you see the whole picture, not just the flags.
Will it give the same result twice?
Yes, and you can prove it. The analysis is deterministic, so the same documents and the same engine version always produce the same report — and the report carries a digest you can verify. It only changes when your project changes or you resolve an open question, never because the tool drifted.
What if it overstates risks just to sell reports?
It can't. Every finding has to trace to your documents, so an inflated or invented risk would fail its own evidence check, and the engine is deterministic — it can't quietly raise severity for one customer. If your project is in good shape, the report says so. A clean, well-evidenced verdict is a valid result, not a failed sale.
Why should I pay if the analysis is fast?
You're paying for the answer, not the runtime. The method — what a sound project must evidence — is pre-built, so applying it is fast; the depth (23 review areas, a full finding register, an appendix) is the same every time. The report's job is to catch the one expensive problem before your decision is made — that's worth far more than the minutes it takes.
Couldn't I just ask ChatGPT myself?
You can, and for a quick gut-check it's fine. But ChatGPT won't give you a traceable, reproducible, board-forwardable assurance report — it gives a different answer each time, cites nothing you can verify, and is poor at spotting what's missing from your project. Project Intelligence is built to produce an artifact you can stake a decision on and defend afterwards.
How is this different from project management software?
Your PM tool tracks the work you already know about. Project Intelligence independently judges whether the project is set up to succeed — surfacing what's missing (an untested rollback, an unfunded change budget) and where your own documents contradict each other. Different job: tracking vs. independent assurance.
How is this different from consultants?
Use both. A consultant gives one in-depth, expensive opinion at one point in time; Project Intelligence gives a fast, repeatable independent baseline. Run the report first to see where the project stands, then focus your consultant's hours where they matter most — and re-run the report whenever your documents change.
Can this replace my project manager?
No, and it isn't meant to. Your PM is the best person to run the project and — because they're inside it — the worst placed to independently review it. Project Intelligence is the outside read: it surfaces what the team hasn't articulated and gives your PM an independent challenge to take to the decision body.
Is this for small businesses?
Yes — there's a version sized for small projects. The report is calibrated to your project's scale, so a lean team gets a proportionate read, not an enterprise audit. The fundamentals (sponsor, budget, rollback, testing) matter at every size, and a small project often has less slack to absorb a missed risk.
Is this for large organizations?
Yes. There's an enterprise example built on a complex programme: it handles many documents across workstreams, produces a board-ready report (cover, one-page glance, finding register, appendix, independence statement), and ties every finding back to source.
What happens to my documents?
Your documents are processed only to produce your report, by a named person under a confidentiality agreement, and are deleted on request — with the terms set out in writing before you share anything. See the Security & data handling page for the full detail.
Are my documents used to train AI models?
No — your documents are never used to train any AI model. The findings come from a deterministic rule engine, not a learning model. One nuance, stated plainly: the optional semantic-retrieval layer sends short document passages to OpenAI to compute embeddings (this is disclosed in every report and on the Security page, and can be turned off). Those passages are processed to generate the embedding, not retained or used for training. See the Security page for exactly how your documents are handled.
What do I get for free?
A real preview from your own documents: how many findings were found, the verdict, the named red flags, and the key unknowns — with the full report transparently held back. It follows the same rule as the paid report: it never invents or exaggerates a risk, and if your project is in good shape, it says so.
What do I pay for?
The full, forwardable assurance report: the verdict, the must-not-fail gate, the failure patterns, the blind spots, counterarguments, a premortem, the critical missing information, prioritized recommendations, decision confidence, and every finding traceable — as a PDF you can forward to your decision body.
What is a project licence?
An optional way to keep the review live: re-run it whenever your documents change, resolve decision-critical unknowns over time, and track how the risk picture moves across the project's life — rather than a single one-off read.
What if the report finds nothing serious?
Then it tells you that, with the evidence — and you've bought a defensible, independent, traceable clean bill of health for your decision record. A reviewer that always finds disasters is selling fear; one that can honestly return a clean verdict is selling assurance.
What does it cost?
A single project assurance report is a one-time fixed fee. The exact price is shown before you unlock the full report — you only pay after you have seen a free preview generated from your own documents. Pricing is still being validated with real customers, and founding customers get founding-customer pricing and direct founder access in exchange for structured feedback.
Still have a question? hello@knowision.com · Security & data handling → · Run a free preview →