Nine capabilities. One accumulating system.
Each capability below runs independently or as part of the full loop - start where the pain is highest and let the system grow into the rest. Every one of them compounds with usage.
Ticket solving
A ticket board that runs the whole loop: file → solve → review → resolve, with a live step-by-step activity trail and an always-available conversation to guide or approve the solve.
Playbooks
Versioned, evolving how-to-solve runbooks per recurring task type - with router triggers, changelogs, and a maturity ladder (draft → tested → automatable) that is the automation roadmap.
Knowledge library
A permissioned library organized by folder → module → content. Hybrid keyword + semantic retrieval, auto-categorization of uploaded documents, and scope-gating on everything it returns.
Approvals
When Kairo proposes a playbook change or a query to run, it lands in Approvals for someone with write access - shown as a GitHub-style diff, with a stale badge if the playbook moved on.
People & access
Onboard a person, assign modules (role/team/domain/product), and they inherit that scope's knowledge. Roles and levels gate what each member sees - manager content stays manager-only.
Tool execution (MCP)
Register real MCP tools - a read-only analytics warehouse, Sheets, an HTTP action - and grant each module a mode: draft only, needs approval, or auto-run. Writes are never auto-executed.
Search
Search across tickets, playbooks, and knowledge - results grouped by type, filterable, clickable, and scope-gated to your modules. Try it with a phrase like 'autodebit mandate failure'.
Dashboards
What needs attention, how the system is learning, and per-solve metrics - tokens, time, tool calls, redundant work saved. Measured on every run, visible everywhere.
Plug-and-play LLM
Any OpenAI-compatible endpoint - bring your own model and key, encrypted at rest, configured per organization in Settings. Swappable embeddings too.
One wrapped system, not a bag of features.
Every capability talks to the same interfaces - a single model layer, a single retrieval layer, a single permission model. That's why onboarding is inheritance: assign the modules for a role and a new hire gets everything the system already knows about that work.
See the loop end to end →Ready to put your recurring work on autopilot?
Join the waitlist and be among the first teams to run Kairo inside their own operations.