Questions
The things people ask, answered plainly
Grouped by what you are actually worried about: what this is, what happens to your data, whether the model can be trusted, what it costs, running it yourself, and the questions that are uncomfortable to answer.
What it can and cannot tell youWhat is built and what is not
What it is
What does this actually do?
It reads the documents you already have, builds a model of your company from them, gives every customer, competitor, supplier, salesperson, executive, employee, investor, regulator and partner its own objectives, and then runs the year forward hundreds of times with different draws from the uncertainty in your own numbers.
What comes back is a median with a tenth and a ninetieth percentile around it, a list of what the agents did and how often, and a brief naming the assumptions the answer is standing on.
How is this different from a financial model in a spreadsheet?
A spreadsheet gives you one path and has nobody in it. It has no competitor who matches part of your move in month four, no salesperson who starts discounting when they are behind, and no customer whose contract happens to end in the month you send the price letter. Those are the reasons a price rise rarely goes the way the spreadsheet said, and they are what this adds.
Is this a language model with a chart on it?
No. The simulation is arithmetic and runs with no API key at all. Given a seed it produces the same answer every time.
A language model does three narrow jobs at the edges, all optional: reading prose the rules could not parse, rewriting a finished brief into better English, and answering questions in Ask. None of them touches the simulation.
Your data
Where do my files go?
Into a folder inside the WordPress uploads directory, outside the media library, under randomised filenames, with a deny rule written into the folder so it is not served over the web. They are never registered as attachments and they are never transmitted anywhere.
Do you train on my data?
No, and there is no pipeline in the product that could. Nothing you upload is used to train, fine tune or evaluate any model, nothing is shared between accounts, and no aggregate or benchmark is built from customer data. The industry priors are fixed values in a file in the source code, not learned from anybody.
What happens when I delete something?
Deleting a source unlinks the file from disk and deletes its chunks and its facts in the same operation. No archive, no trash, no hidden flag. Deleting a twin takes its sources, chunks, facts, assumptions, scenarios, runs and questions with it.
What leaves the server?
Nothing, unless a language model key is configured. With one, three things can leave: chunk text during extraction, a finished analysis when you ask for the brief to be rewritten, and your question plus its retrieved context in Ask. No file is ever transmitted as a file.
The model
How accurate is it?
Its accuracy is the accuracy of its assumptions, which is why they are all listed with bands and origins. It is good at shape, order of magnitude, timing, which actor causes the damage and how wide the uncertainty is. It is bad at predicting a market, at one off events, at anything your files do not contain, and at long horizons.
There is a whole page on this, and it is worth reading before you pay for anything.
What if I do not have good data?
Then the model says so. It runs on industry defaults and the readiness score tells you how much of the answer is standing on them. The missing files are listed in the order it is worth fixing them. A twin built from one profit and loss and a customer list is already more useful than an argument.
How fast is it?
A standard run is 200 replications of the scenario and 200 of the baseline, which is 400 simulations, and takes about a tenth of a second for a twelve month horizon with around forty agents on the machine it was built on. A 25 step sweep at 80 replications is 2,000 simulations.
Will I get the same answer tomorrow?
Yes. The seed is stored with the run, so the same twin, the same levers and the same seed produce identical numbers. That is what makes it possible to change one assumption and know that what moved was the assumption.
Has it been validated against real companies?
No. No backtest against known outcomes has been done or published. When it is, it will be on the changelog with the numbers, including the cases where the model was wrong. Until then the calibration is untested and this site says so everywhere it matters.
The money
What does it cost?
Nothing, if ten runs a month is enough. Above that, Explore is 149 a month, Operator 390, Boardroom 890, and Enterprise is a conversation. Yearly is ten months for the price of twelve. Pricing is not settled and may change before general release.
Is there a free tier?
Yes, since version two. One twin, three files, ten runs a month, twelve month horizon, sixty replications. No card and no expiry. Everything that makes the output honest is in it: the full ledger, every origin, every band and every Why chain. What is missing is scale and the network. The pricing page explains why the earlier answer to this question was no, and why that was wrong.
What happens to my twins if I stop paying?
You drop to the free tier, so one twin stays open and readable. Anything above that limit is kept for thirty days so a lapsed card or a cancellation you regret does not destroy your work, and after that it is deleted rather than archived. Export before you go.
Running it yourself
Can I run this on my own server?
Yes. It is two WordPress plugins: one is the application, one is this website. Install them on your own WordPress and nothing leaves your building unless you configure a language model key yourself.
What does it need?
A reasonably current WordPress on a reasonably current PHP, with MySQL or MariaDB. The ZipArchive extension is needed for Word and Excel files and is present on most hosting. Nothing else, no queue, no separate service, no node process.
Is there an API?
There is a REST namespace that the application itself uses, authenticated as a signed in WordPress user. There are no API tokens for third party use yet, which means automating this from outside the browser is not supported today. It is on the roadmap and it is listed as not built on the status page.
Can I see the source?
If you self host, you have the source: it is PHP in a plugin folder, not compiled and not obfuscated. Read the extraction rules, read the engine, and check that the claims on this site are true. That is a deliberate property rather than an accident.
The uncomfortable ones
How many customers do you have?
None to point at. This is new software and there are no logos, no testimonials and no usage statistics on this site, because inventing them is the easiest lie in this industry and the whole site is written to be checkable instead. When there are customers willing to be named, they will appear here with their permission and nothing sooner.
Who built this?
One person. There is no team, no funding round and no advisory board, and the about page says the same thing in more detail. That is a real risk to weigh, and it also means the person you email is the person who wrote the code.
Has this run in production anywhere?
Neither plugin has run on a production WordPress install yet. The two things genuinely unknown are how the table creation behaves against a particular MySQL configuration and whether the payment wiring works end to end with a real card. Both are named on the status page rather than discovered by you.
What is the most likely way this misleads me?
A confident median with a very wide band around it, read by somebody who only quotes the median. Or a beautifully run simulation of a question that is not the decision you are actually facing.
Both are covered on the accuracy page, along with three more failure modes and four habits that avoid them.
Why should I trust any of this?
Do not trust it. Check it. Every number in a twin points at the line in your file it came from, every assumption is listed with its origin, the build status page says what is not built, and if you self host you can read the code that does all of it.
Still have one that is not here?
Ask it. The answer goes to you first and, if other people are likely to ask the same thing, onto this page afterwards.