Agent Library · By platform · Comms · HITL console · Alerts
L
Where approving an agent takes one tap.
Slack matters twice in a Kamaan build. It is a source of work to be routed, and it is the most practical place to put the human-in-the-loop console — because approval that requires opening another tool is approval that does not happen.
Decisions get made and lost
The thread that settled it scrolls away and is re-litigated a month later.
Channels outpace attention
Being in forty channels means genuinely reading none of them.
Requests arrive as messages
Real work enters as a DM and leaves no trace in any system.
The agents
5 agents. Each one earns its autonomy.
Agent
What it does
Autonomy
Human gate
HITL Console
Every pending agent action surfaces here with the proposed output, confidence and evidence. Approve, edit or reject.
Review
This is the human gate
Channel Summariser
Catches you up on what actually changed in the channels you missed, not everything that was said.
Auto
—
Request Router
Recognises real requests inside conversation and files them into the system that should own them.
Notify
—
Decision Capturer
Detects decisions in threads and records them with context, owner and date.
Review
—
Alert Consolidator
Correlates and de-duplicates automated alerts so the signal survives the volume.
Auto
—
Autonomy key — Auto: executes above 95% confidence. Notify: executes at 85–95% with notification and undo. Review: always queued for a human. Flag only: never acts, alerts a person.
What changes
Three shifts, not a feature list.
Approval takes one tap, so it actually happens
The single biggest failure mode of human-in-the-loop design is friction. This removes it.
Decisions stop evaporating
What was settled gets recorded where it can be found.
Alert fatigue reverses
Consolidation restores the meaning of an alert.
The eval bench
What these agents have to pass before they touch real work.
- Every approval action logged with approver, timestamp and the exact output approved
- Decision-capture recall measured against human-identified decisions in the same 50 threads
- Zero agent action executed from Slack without an auditable approval record
Written against your data, not ours. The bench is built during Blueprint and runs on every deploy from then on.
Related
Where this connects.
See it on your Slack.
A Blueprint maps which of these agents pays off first in your exact setup — and which to leave alone.
Apply for a Blueprint