About
Counterfirm was built by Lewis De La Paz. One person, no team, no funding, no customers to point at yet. This page says that plainly because the rest of the site would not be worth reading if this one were dressed up.
No invented team. No invented milestones. No logos.
Every company runs on decisions that nobody can answer properly. Raise the price or hold it. Keep the department or cut it. Take the acquisition or pass. The way these get decided is a spreadsheet with one path through it, a meeting where the most confident voice wins, and a consultant six weeks later with a deck.
The spreadsheet is the interesting failure. It is not wrong because the arithmetic is wrong. It is wrong because it has nobody in it. There is no competitor who matches part of your move in month four, no salesperson who starts discounting when they are behind on quota, no customer whose contract happens to end in the same month your price letter lands, and no executive who panics in month two and undoes the whole thing.
Simulation with agents in it is not new. It has been used in epidemiology, in traffic, in markets and in military planning for decades. What has been missing is a version an operator can actually use: one that reads the files a company already has, builds the actors itself, and returns an answer with the assumptions attached rather than a research project.
That is what this is. Whether it is good enough is a question for people who have used it on real decisions, and there are not any yet.
The person
The software, the engine, the extraction rules, the agent generation, the report writer and this website were all written by one person. When you email support, that is who replies. When something breaks, that is who fixes it.
The honest version of what that means: decisions get made quickly and there is nobody between you and the person who can change the code. It also means there is no redundancy, no on call rotation, and no second opinion inside the building. Weigh that properly before putting this in a critical path, and if you need the resilience of a larger company, this is not one.
How this site is written
Every rule below cost something in persuasiveness. They are here because a product that asks for your financial statements has to be worth more than its marketing.
There are no customers yet. A quote from a satisfied chief executive would be invented, and a chart of runs per week would be invented, so neither exists. When there are real ones they will be attributed to real people who agreed.
Harborline Components is a contract manufacturer with 24 customers, 14.24 million dollars of revenue and one account at 18 percent. It is invented, it is described as invented every time it appears, and it exists so the screenshots problem is solved without pretending.
You will not find a sentence saying companies using this see a percentage improvement in anything. No study is cited. The industry priors inside the product are described as starting points, because that is what they are.
The build status page names every capability the site mentions and marks it built, partly built or not built, including the two things that have never been tested in production. Most products do not have that page. This one does, and the refund policy is tied to it.
What it can and cannot tell you is written for somebody who has not paid, and it is linked from the footer of every page. If a marketing claim elsewhere contradicts it, that page is right.
Nine agent classes, 24 levers, 15 scenario templates, 400 simulations in a standard run. Each of those is a count you can verify in the source if you self host, and they are kept the same on every page on purpose.
There is no social proof on this site, and that is the point.
A new product with no customers has two options: borrow credibility it has not earned, or spend the effort making everything checkable. The second is slower and works better on the kind of person who would be careful about uploading their accounts to unfamiliar software. If you are that person, start with the build status page, then the accuracy page, then decide.
Three things, none of which exist yet, all of which will be announced on the changelog when they do.
Until then this is a piece of software you can inspect, run on your own hardware, and test against a decision whose outcome you already know. That is a reasonable basis for a judgement, and it is the only honest one available right now.
Bring one decision and one file. If the twin it builds is useful on that decision, the rest is worth your attention. If it is not, no amount of vision makes up for it.