What it costs, and who is liable

What it costs

Nothing, and it is not for sale. There is no price, no trial, no order form and no licence today.

That is not modesty and it is not a launch tease. Three things have to be settled before there could be a price, and none of them is:

• the licence — the repository's licence file and its notice file state different scopes, and reconciling them is the founder's decision, not a drafting exercise. Nothing may be offered under any terms until it is settled, because a grant cannot be withdrawn from a copy already received;
• the build — nothing has ever been observed to produce the same answer on a second interpreter or a second operating system, and this project's own rule is that nothing ships until it has;
• the liability position — there is no professional indemnity insurance, no cyber liability insurance, and no executed agreement of any kind.

If it later becomes a product, three things are already settled and can be said:

• It would be sold as a tool, not as a service or a study. Not engineering, not consulting, not an evaluation, not a study, not program management. That discipline is not marketing: it is simultaneously the professional-licensure defence, the qualifications-based-selection avoidance strategy and the insurance-exposure control. One positioning decision solves three unrelated legal problems, and it is the same decision as supplies a computation, does not supply an opinion.
• It can be self-hosted. The engine is pure standard library with no third-party runtime dependencies at all, enforced as a build failure, so it runs inside your own environment with no network access. That matters for whether cloud-authorisation regimes are even in scope.
• The price would be set below a threshold a purchasing office can actually use, rather than against the size of the dispute avoided.

That last sentence is the discipline and it is worth stating in full, because the temptation runs the other way. A large average dispute value does not mean a subscription prevents one. Projects running full P6 with quantified schedule risk analysis have gone billions over. The product verifies and records; it does not improve outcomes, and pricing that implies otherwise would be challenged by exactly the sophisticated buyer worth having. Price against the cost of the work displaced, never against the size of the disaster avoided.

Will this become something you sell to me later?

Possibly, and it would be dishonest to say otherwise. There is no product today and no plan being executed against your office. If it becomes one, you will be told plainly that the conversation has changed character.

Who is liable

Nobody has signed anything, so this section describes a posture rather than an agreement. Read it as what the agreement would have to say.

The tool is built so that it is not the thing being relied on. It does not conclude. On the questions that decide a dispute — is this delay excusable, who is responsible, is there concurrency — it either refuses outright or returns a range with both theories named. A conformance finding is "activity 9200 has no predecessor, which §3.3.11 does not permit", which is checkable by anyone with the file open.

It is deterministic, with no model in the answer path. Same input, same version, same output, and the version is printed on the report. Reproducibility is separately a Daubert factor, and it is why there is no language model anywhere in the computation.

It is not validated at scale, and a CPM arithmetic error is possible. If this engine computes a float value P6 would not, and someone relies on it, that is a real failure mode with real consequences. Any figure the tool produces should be reconciled against the source scheduling software before it is relied on — the report says so itself, on every run.

There is no professional indemnity insurance today, and no engineer's seal. Nothing here should be presented as an expert opinion, because it is not one, and no software output is a substitute for the expert a party retains. Every report carries the authorship statement in its own words: "Produced by software. No professional engineer, licensed under any state practice act, has reviewed or adopted this output."

If that combination is disqualifying for your purpose, it should be. The correct use of this tool today is to find the mechanical defects before a person spends an afternoon on them — not to decide anything.

Four things any future agreement would have to survive

Named here rather than discovered later:

1. Reliance by somebody who was not the customer. The distinction to draft is between reliance on the output, which no third party gets, and the ability to obtain and run the software to test the output, which the opposing expert must have or the whole disclosure argument collapses. Those pull in opposite directions and both are load-bearing.
2. A wrong input supplied by the customer. A wrong contract term produces a confident wrong finding with no signal in the report. The agreement has to place responsibility for supplied terms on the customer explicitly rather than by implication.
3. A misread specification. What the software could warrant is that it applies the stated rules and reports the result — and nothing about whether a rule is a correct reading of the clause. Every finding cites an external published authority, and this project never asserts its own standard of good practice.
4. Gross negligence and wilful misconduct. Most jurisdictions will not let a liability cap reach them. That is conceded early rather than defended.

One question that is genuinely unresolved

Does a reservation of determination actually reserve the determination, or does a report saying "fails §3.3.11" make one regardless of what the footer says?

That is the load-bearing open question on this whole subject and it is for a solicitor with both software licensing and construction disputes on their record. It is recorded here rather than answered because nobody here can answer it.

What this project does not have

Said plainly, because a promise nobody can audit is worse than no promise:

No SOC 2. No ISO 27001. No FedRAMP. No penetration test. No cyber liability insurance. No professional indemnity insurance. No certifications, accreditations, endorsements or reference customers of any kind. No customer, no pilot, no agency. The author holds no PE licence, no PSP and no CFCC.

If your office requires any of those before a file can be shared, it does not meet the bar today — and it would rather you applied that bar than waived it.

What data would be handled, and how

Decided, though untested by any real deployment:

• The tool runs locally and sends nothing anywhere. No network calls. A file processed on your machine does not leave it. There is no cloud service to breach because there is no cloud service.
• Should there ever be a hosted surface: United States only, no non-US processing, no non-US subprocessors, no customer data in any model call, and no cross-customer analysis of any kind by default.
• Not a multi-tenant upload service for live-dispute files, and that is a decision rather than a gap. A consultant who signs up with a card and uploads an opposing party's native schedule to a service they chose has made a disclosure to a third party outside the tent, in a browser, at eleven at night — and an archive that may never be deleted grows monotonically and stays subpoenable for as long as the company exists.

If you have a file you want looked at

Where a project is in dispute or a dispute is anticipated, route it through your counsel of record rather than here. Analysis produced outside a litigation instruction may not attract work-product protection, and a file handed to a third party is a disclosure your opponent can ask about.

"Completed" is not the test. A project can be finished and its claim still live. The question is whether a dispute exists or is anticipated, not whether the work is done.

Source: web/pages/pricing.md. Source commit date: 2026-09-06.