integrations Crisp
Every chat, read. Cogny turns support into briefs.
Cogny reads Crisp and makes the change you approve, through 16 tools, from your own AI or the Cogny app. Then it measures the change in your data and gives it a verdict.
what it can touch.
- 11reads
- Get a website's metadata.
- List operators (team) on a website.
- Get overall website statistics.
- List conversations.
- 5changes
- Send a message in a conversation.
- Update conversation state: pending | unresolved | resolved.
- Create a new conversation (returns session_id).
- Create a people profile.
- 0removes
- none
all 16 tools, with what each does
what it reads.
- CrispConversations, Contacts
- CRMOutcome data
- GA4Site context
Read side by side, so a change in Crisp can be judged by what moved somewhere else.
what it proposes.
- Rising-intentA new question category is showing up; brief for content/product.
- Unresolved-patternRecurring blocker your team hasn't documented.
- Pre-sale-patternQuestions that come up before a buyer converts.
what it replaces.
How a Crisp review is done by hand today:
- Support team manually tags conversations — or doesn't.
- Top FAQs reviewed quarterly, if at all.
- No way to see which conversations converted.
- Product + content miss obvious gaps the chat reveals.
With Cogny you ask once, approve one change, and get a verdict three days later (seven for pages, posts and products).
then a verdict.
- You ask, or a Loop runs on schedule.
- Cogny proposes one change.
- You approve. It ships in Crisp.
- Three days on, seven for pages, posts and products, it is measured.
- worked
- partial
- didn't work
- inconclusive
Misses stay on the record. Every plan, Free included.
questions.
Does Cogny replace Crisp?
No. You keep Crisp. Cogny connects to it over MCP and works on top of it.
Does it change anything without asking?
From your own AI client, that client asks before a write. In the Cogny app, write tools ask first by default, set per tool. Any tool can be blocked in Settings → Integrations, and every write is logged with the tool, time and result.
How do I know a change worked?
A launch (a new campaign, ad or similar) becomes a measured ticket on its own, from your own AI or the Cogny app; for other changes your AI offers to register them. Three days later (seven for pages, posts and products) the ticket is measured in your connected data and closed as worked, partial, didn't work or inconclusive.
Which plan do I need?
Any. Free includes 15 MCP calls a month, no card. Solo is $9/mo with MCP unmetered under fair use. Cloud, $499/mo, adds a team workspace, scheduled reports and generated tickets.
Which AI clients can use it?
Claude (Claude Code or a Claude.ai connector), ChatGPT, Cursor and Codex, or the Cogny app. In Claude Code it is one line: claude mcp add --transport http cogny https://app.cogny.com/mcp, then authorise in the browser and connect Crisp in Cogny.
Can it run every week on its own?
Yes. On Solo, Cogny AI and Cogny Pro, a finished chat can become a Loop that reruns on a schedule, and each run is its own measured ticket. Cloud adds scheduled reports and automations for a team.
What can't it do in Crisp?
The tool list shows every action it can take. Anything not on it, your agent can ask for with request_feature.
pair it with.
see a Crisp demo on your data.
For teams on Cloud. Tell us what you want done in Crisp; we come to the call with it running on your data.
On your own? Start free, no card. · 16 tools