Generate content, correct spelling, suggest synonyms to enhance writing.
173M visits/mo+0.1%#1 Productivity & Office
$6Starting price
About Kaiku
Kaiku combines a task tracker and a wiki designed for teams that use AI agents as part of their workflow. Agents can be called directly in issue comments or via MCP, performing reads and proposing actions that require human approval. The platform records every agent run against the specific issue, tracking tokens used, machine time, and the user who initiated the request, providing visibility into automation costs. The wiki uses the same storage format as common corporate wikis, allowing existing tools to read and write content without modification. Projects are exported as structured JSON, Markdown, and attachments in a single zip file, ensuring data portability even if the subscription lapses. Kaiku supports importing projects from popular task managers while preserving issue keys, comments, and custom fields. Agents operate under the permissions of the user who invoked them, with no separate accounts or tokens, and their outputs are written by Kaiku rather than the model itself. The system emphasizes transparency by including the sources and costs of every agent interaction within the issue thread.
Key features
- MCP server integration for agent communication
- Task tracker with issue and sub-task management
- Wiki with Markdown support and Word document export
- Agent call-in via comments or direct invocation
- Cost tracking per agent run (tokens, machine time)
- Project import from popular task managers
- Structured JSON and Markdown export for data portability
- User-permission-based agent access control
Use cases
- Teams using AI agents for task automation within issue tracking
- Organizations needing a wiki that integrates with existing corporate tools
- Projects requiring transparent tracking of agent actions and costs
Pros
- Agents integrate directly into existing workflows via MCP without requiring additional bridges
- Tracks agent usage and costs per issue, enabling cost monitoring and accountability
- Wiki uses standard storage formats, ensuring compatibility with existing corporate tools
- Data export includes structured JSON, Markdown, and attachments for portability
- Agents operate under user permissions, preventing unauthorized access to unrelated projects
Cons
- No free tier available; requires a subscription
- Limited to MCP-based integration, excluding other APIs or plugins
- Agent actions require human approval, limiting fully autonomous operation
- Data export functionality may not cover all historical changes or sprint data
Frequently asked questions about Kaiku
What is Kaiku and who is it designed for?
Kaiku is an agent-native task tracker and wiki designed for teams that incorporate AI agents into their workflow. It is built for organizations that want agents to operate alongside human team members, performing tasks with human approval and maintaining transparency in automation costs.
How do agents interact with issues in Kaiku?
Agents can be called directly in issue comments or via MCP, where they read the issue, propose actions, and provide reasoning based on the content they accessed. Their outputs are written by Kaiku, not the model, and require human approval before any changes are made.
Does Kaiku support importing existing projects from other task managers?
Yes, Kaiku allows importing projects from popular task managers while preserving issue keys, comments, custom fields, and other metadata. A pre-import check identifies what will not transfer, such as change history or logged time.
How does Kaiku handle data portability and ownership?
Projects can be exported as a single zip file containing structured JSON, Markdown, and attachments, ensuring data remains accessible even if the subscription lapses. The wiki uses a storage format compatible with common corporate wikis, allowing existing tools to read and write content without modification.
What permissions do agents have in Kaiku?
Agents operate under the permissions of the user who invoked them and do not have separate accounts or tokens. They are restricted to the project of the issue they are working on, and their tools are read-only, with outputs written by Kaiku rather than the model itself.
How does Kaiku track automation costs?
Every agent run is recorded against the specific issue, including tokens used, machine time, and the user who initiated the request. This provides visibility into the actual cost of automation while it is still possible to take action.
