Sayli is built for people in real conversations. This policy draws the line around that. It is part of the Terms of Service.
Do not
- Break the law, or use Sayli to help someone else break it.
- Record or transcribe a conversation without the consent the law requires. See Recording consent guidance.
- Upload content you have no right to use, or malware of any kind.
- Probe, scan, or try to defeat the service's security or another organization's tenant isolation.
- Scrape the service, or resell or sublicense access to it.
- Pass your own use off as somebody else's, or somebody else's as yours: one account serving several people, sharing a login, reselling capacity, or running a script that fabricates the volume of a person who is not there. Using the API, a personal access token, or the MCP server from your own code and your own AI tools is not this. Those exist to be used.
- Use share links, invite emails, or the referral program to send spam, or farm referrals with fake accounts or devices.
- Impersonate someone, or present Sayli's output as a human's verbatim words where that deceives.
- Use Sayli to gain an undisclosed advantage in an examination, assessment, technical screen, interview, or certification, or anywhere else that use of AI is prohibited or has to be declared.
- Present Sayli's output as your own work where that is prohibited or would deceive the person receiving it.
- Use Sayli in a way that breaks your employment agreement, your school's rules, or the terms of the platform you are using it on.
What "unlimited, fair use" means
Paid plans include unlimited live copilot and unlimited asks. Concretely:
- There is no monthly ceiling on meeting minutes, no cap on asks, and no overage charge, for a person using Sayli in their own calls and their own workspace. Heavy daily human use is the intended use and will never hit a fair-use limit.
- Two operating limits do apply to a single session, on every plan including paid, and neither is a fair-use judgement about you. One meeting records up to 4 hours of live audio, counted across reconnects, and Sayli warns you five minutes before. After that, starting the same meeting again is refused and you start a new one. A session also stops after 20 minutes with no speech, again with a warning first, so a call you walked away from does not record an empty room. Neither limit deletes anything already captured.
- One more ceiling exists for runaway scripts, not people: a daily safety limit on AI answers, on every plan, set well above the heaviest honest day. If a paid workspace ever reaches it, Sayli pauses AI answers until midnight UTC and tells you so. Email support@sayli.ai and we lift it.
- We publish the one number so "unlimited" is not a word game: 60 hours of live copilot a month is the point at which an account becomes worth a human look. It is not a cap. Nothing blocks, throttles, or bills at that line.
- Per-minute rate limits protect the service for everyone. A person typing questions never reaches them; a script does.
- Fair use ends where automation begins: bots or scripts issuing asks, one account serving many people, account sharing, or reselling capacity.
- Every plan is also subject to the Knowledge Base storage limits published at Pricing and plans, and we give at least 30 days notice before reducing any published limit.
If usage looks automated, we contact you first unless it is plainly abusive. If it continues, we apply temporary limits, and as a last step suspend the account.
Enforcement
We may remove content, limit, suspend, or terminate accounts that violate this policy, as described in the Terms of Service.
Reporting
Report abuse to support@sayli.ai. Report security issues to security@sayli.ai.