Visual trigger-action editor. Pick a trigger (a ticket arrives, a bug is resolved, a churn signal fires), describe the work, connect the outputs. No engineers required.
Runs on the right data
Tickets, bug reports, feature requests, NPS, in-app events, customer tier - your custom agents see what Gleap's built-in agents see. The whole customer story, not just a CRM row.
Three data layers
Gleap data + your business systems + 1,000+ external integrations.
MCP server included
Build, test, and run agents from Claude, Cursor, VS Code.
Composio under the hood
1,000+ integrations available out of the box.
SOC 2 + audit logs
Every agent action logged. Approval gates available.
Three layers of access
Gleap data + your business systems + the whole internet.
Gleap data, your business systems (Salesforce, Stripe, Snowflake), and 1,000+ apps via Composio. If your team uses it, an agent can read it, write to it, or trigger it.
OpenAI GPT, Anthropic Claude, Google Gemini, xAI Grok, and open-source model families like Llama, Mistral, Qwen, and DeepSeek - switchable per agent. Or connect your own model endpoint on Enterprise for full control over cost, latency, and data residency.
Partfox fully integrated Gleap with custom data in 10 days and now uses live chat, AI support, bug reporting, feature requests, product tours, checklists, and surveys.
Maic Fankhauser
Product Manager, Partfox
Agents teams ship in an afternoon
Six real examples. All possible today.
01
The Churn Defender
Sentiment drops below threshold over 14 days → agent reads full history, drafts personalized outreach, queues for CSM approval with one click. Risk becomes visible.
02
The Revenue Signal
High-ARR support conversation → agent extracts the business need, creates a Salesforce opportunity with the full conversation summary, notifies the account owner in Slack.
03
The Competitor Monitor
Scheduled crawl → agent reads competitor changelogs, pricing pages, app store reviews. Posts a Slack brief weekly. Flags roadmap items that may need acceleration.
What you can connect
Your data, your systems, the whole internet
Gleap platform data
Tickets, bug reports, feature requests, NPS, in-app events, customer tier - the full Gleap data layer.
Kai Custom Agents is Gleap's no-code AI agent builder. It runs on the same platform that powers Kai, Kai Resolve, Kai Code, and Kai PM - with three layers of access: Gleap's customer data, your business systems (Salesforce, Stripe, internal APIs), and 1,000+ external integrations via Composio.
How is Kai Custom Agents different from Zapier?
Zapier is task automation - A triggers B. Kai Custom Agents is AI agents that reason: read context across multiple sources, make a judgement, write personalised output, take an action. Plus they run on Gleap's customer data layer - not just whatever a Zap can scrape. Custom Agents share Gleap's audit logs, approval gates, and SOC 2 posture.
How many integrations are available?
40+ native Gleap integrations + 1,000+ via Composio. Slack, Jira, GitHub, Salesforce, HubSpot, Stripe, Notion, Google Workspace, Mailchimp, Customer.io, Snowflake, BigQuery, and on. If your team uses it, a custom agent can probably connect to it.
Do I need to code?
No. Visual trigger-action editor. Pick a trigger (a ticket arrives, a bug resolves, a churn signal fires), describe the work in natural language, connect the outputs. For technical users, the MCP server lets you build, test, and run agents from Claude, Cursor, or VS Code.
Which models power custom agents?
OpenAI GPT, Anthropic Claude, Google Gemini, xAI Grok, plus open-source model families like Llama, Mistral, Qwen, and DeepSeek - switchable per agent. Enterprise plans support bringing your own LLM endpoint for cost, latency, and data residency control.
Are custom agents auditable?
Yes. Every action is logged. Sensitive actions can be gated behind approval (via the Tasks feature). Full audit log accessible from your Gleap workspace. SOC 2 Type 2 audited. EU residency available on Enterprise.