AI that inherits your permissions.
Most CRM AI is a chat panel wired to a separate pipeline, which is why the honest answer to “what can it see?” is usually “more than that user can”. Unify Loop’s AI is built on the same object and query layer as every screen, so scoping and permissions aren’t re-implemented for it — they’re inherited.
- Permission model
- Inherited, not re-implemented
- Search output
- A reusable filter, not just text
- AI messages on timeline
- Labelled as AI
- Spend controls
- Per-workspace budget ledger
It uses the product, not a copy of your database
The AI works through a registry of tools that read and write via the same layer your UI uses. When it searches, it emits the same filter format a saved view uses. When it updates a record, the same validation and field permissions apply. That has a useful consequence: the AI cannot perform an action the signed-in user isn’t allowed to perform, because there’s no second path for it to take.
Every answer shows its work
Ask “which enterprise deals went quiet this month?” and you get the answer plus the filter it ran — inspectable, correctable, and saveable as a view. Ask it to draft a reply and you see which record fields and which prior messages it drew on. AI you can audit is AI you can eventually let act on its own; AI that returns a confident paragraph from nowhere never earns that.
- Natural-language search that returns a real, reusable filter
- Drafted replies grounded in the record and the thread, labelled as AI
- Summaries of long threads and record histories
- Scoring and prioritisation you can inspect rather than a mystery number
Spend is metered and capped, per workspace
AI usage runs through a budget ledger with per-workspace limits, so cost is visible before it’s a surprise on an invoice. You can see which features consumed what, cap a workspace, and let the platform degrade gracefully at the limit instead of either failing silently or spending without a ceiling.
AI copilot, in detail.
Is my data used to train models?
No. Your workspace data is used to answer your prompts and nothing else. See our security page for the detail on subprocessors and retention.
Can the AI see records a user isn’t allowed to see?
No. It reads through the same query engine that enforces row and field permissions for the UI. There is no separate index with looser scoping, which is the usual reason bolted-on CRM AI leaks.
Can I let AI send messages without a human reviewing them?
Yes, per workflow, within limits you set — and consent and quiet-hour rules still apply. Most teams start with AI drafting for human approval and expand autonomy once they’ve read a few weeks of drafts.
What happens when I hit my AI budget?
AI features degrade with a clear message rather than failing opaquely, and admins are notified before the cap rather than after. The rest of the platform is unaffected.
Try it on your own data.
Free workspace, an industry pack installed at setup, and every price published. No mandatory call.
Free trial · no card required · every price published