Kaiku launches an agent-native task tracker and wiki with wire-level compatibility for existing corporate tracker APIs and native MCP integration.
Kaiku is an agent-native task tracker and wiki built for engineering teams whose workflows increasingly involve AI agents like Claude Code and Cursor. Rather than requiring teams to install fragile bridges or retool existing integrations, Kaiku speaks the wire-level REST API and error formats of incumbent enterprise issue trackers, enabling existing agent MCP servers to function unchanged. Alongside traditional kanban boards, backlogs, and wiki spaces, Kaiku lets users summon agents directly inside issue comments to investigate context and propose actions without unilateral decision-making authority. It also tracks operational telemetry—such as token consumption by type, run time, and total cost—directly on the associated issue tickets.
Treating the issue tracker itself as the native orchestration canvas for AI agents—complete with wire-level API parity and granular cost tracking—is a much smarter tactical wedge than attempting to build yet another isolated agent IDE or dashboard.
* Wire-level API parity: Mimicking the REST request shapes and MCP interfaces of established corporate trackers removes friction, letting existing agent toolchains connect seamlessly.
* Governed agent collaboration: By strictly enforcing a "propose, never decide" pattern and clamping agents to the invoking user's permissions, Kaiku establishes guardrails against autonomous hallucination or unauthorized changes.
* Issue-level cost accounting: Directly attaching token usage and dollar costs to tickets solves a major enterprise pain point around runaway LLM spend during automated software development.
* Entrenchment challenges: Despite clean agent primitives, displacing deeply entrenched project trackers like Jira or Linear in established companies remains a steep organizational barrier.
DISCOVERED
1d ago
2026-09-25
PUBLISHED
1d ago
2026-09-25
RELEVANCE
AUTHOR
Stanislav