In effect · 12 August 2026 · the short version
The plain-language version of what our output is and is not. It is a warning, not a contract — it describes how the software behaves, which is true today. The full Terms of Service are still in draft.
WeYala plans, designs, writes and analyses using software that makes probabilistic judgements. It is not following a script and it is not looking up a right answer. It produces the most plausible result it can, which is usually good and is sometimes confidently wrong.
That is a property of the technology, not a defect we expect to remove.
Read the output before you use it, and have someone qualified read anything that matters. Do not deploy generated code, act on a generated financial model, send a generated message, or make a decision that would be expensive to reverse on WeYala's word alone.
WeYala tries to tell you how sure it is — a stated accuracy, a comparison against a simple baseline, a flag that a number is estimated rather than measured. That exists so you can judge the work.
It does not turn the work into a guarantee. A confidence score is a claim about the method, not a promise about the outcome.
By design, not by omission:
Where a decision is genuinely consequential — architecture, scope, anything irreversible, anything touching multiple users' data — it stops and asks you.
What WeYala produces for your project belongs to you, and you can take it and run it anywhere, including after you stop using WeYala. Two honest limits: work generated by software may not attract copyright everywhere, and we cannot promise that it does.
Privacy Policy — what we collect and who else sees it (in effect).
Sub-processors — the providers we use and their own policies (in effect).
Terms of Service — the full agreement (draft, not in effect).
Questions: founder@weyala.ai.