Cited answers
Ask the project. Every answer shows its sources.
Ask what was agreed, who owns what or why something changed. Velkri answers from the project’s chat, calls, files and email, and every claim carries a footnote you can open. If it cannot cite it, it does not say it.
How it works
Cited answers in three steps.
- 01
Ask in plain words
Ask privately, or mention Velkri in the channel so the whole team sees the answer.
- 02
Velkri reads only what you can open
Answers are built only from sources the person asking has access to. Direct messages are never used.
- 03
Open any footnote
Each citation opens the exact message, file or moment in the call that the claim came from.
What you get
The details that matter.
- Answers only from what you are allowed to see
- Footnotes that open the exact message or moment
- Ask privately or in the channel
- Answers from chat, calls, files and email
Questions
About cited answers.
Who can see what Velkri remembers?
Only the people who could already see the source. Projects are private to their members, answers are built only from sources the person asking has access to, and direct messages are never used for AI or project memory. Access is enforced in the database, not just in the app.
Can we use our own AI model?
On the Enterprise plan, admins can run Velkri on their own OpenAI-compatible endpoint, such as Azure OpenAI, vLLM or LiteLLM. Velkri checks the endpoint and runs each job on a model that can do it. On every other plan Velkri runs the AI for you, through providers that keep no copy of your data and never train on it; meeting bot transcription is the one exception, explained on the security page.