CLI Field Notes Meeting source

Kos Team meeting / October 5, 2026

Cool ideas should become team capability.

The team discussed using GitHub Copilot CLI as both a discovery interface and a contribution mechanism for shared AI practices. The goal: stop useful techniques from disappearing after the meeting ends.

Explore the ideas down
Verified discussion 6 reusable concepts captured 1 operating model inferred
CaptureRecord useful techniques where they are discussed
OrganizeTurn demonstrations into repeatable guidance
DiscoverLet CLI search the team's accumulated learning
ReuseMove code and practices into shared team spaces

WHAT THE TEAM DISCUSSED

Six ideas worth carrying forward

These cards summarize direct meeting evidence. They do not claim that the full automation or every example has already been implemented.

01 Direct evidence
KB

Search a shared "What's Worked" library

Point GitHub Copilot CLI at the team's accumulated techniques, memory setups, project patterns, and best practices before starting a new problem.

Why it mattersReuse proven approaches instead of rediscovering them.
02 Direct evidence
BI

Query Kusto and Power BI in natural language

Ask for case distributions, representative counts, P50 or P90 values, and other operational metrics without first knowing the correct KQL, tables, or reports.

Why it mattersShorten the path from a business question to usable evidence.
03 Kos proposal
QL

Create an AI PM "quest list"

Maintain a growing catalog of capabilities every AI PM can learn, such as querying data through CLI, creating AI-assisted videos, and documenting reusable workflows.

Why it mattersMake skill development visible, practical, and cumulative.
04 Kos proposal
5M

Use a five-minute meeting contribution

Let one person demonstrate a useful technique during the regular team meeting, then capture it while the context, examples, and questions are still fresh.

Why it mattersAttach documentation to existing team behavior.
05 Agreed direction
GH

Move reusable work into team GitHub spaces

Store reusable code and project assets in shared team repositories rather than relying on personal repositories and one-off access grants.

Why it mattersImprove continuity, access, and reuse across projects.
06 Demonstrated example
RP

Turn fast prototypes into reusable playbooks

The Room Pulse example showed how quickly an idea can become a working prototype. The next value comes from sharing the complete process, not only the final demo.

Why it mattersMake the method portable, not just the artifact impressive.

THE INFERRED OPERATING MODEL

A self-reinforcing team learning loop

This model was not stated as one formal architecture, but it is strongly supported by the discussion: CLI helps the team contribute knowledge, and that shared knowledge makes CLI more useful for the next person and project.

Important constraint Allison noted that permission changes and extraction reliability can break fully automatic workflows. The loop needs explicit failure handling and source checks.
01Discover

A team member finds a useful CLI workflow.

+
02Share

The technique is demonstrated in a team meeting.

+
03Capture

Evidence becomes a concise, source-linked how-to.

+
04Reuse

CLI recommends the learning for a future problem.

NEXT PRACTICAL EXPERIMENT

Start with one meeting, one idea, one reusable page.

  1. 1

    Select one demonstrated CLI workflow with enough supporting detail.

  2. 2

    Draft a source-linked how-to with prerequisites, steps, and limitations.

  3. 3

    Publish it in the shared knowledge area and record a stable identifier.

  4. 4

    Test whether another team member can find and reuse it through CLI.

SOURCE AND TRUST

Grounded in the Kos Team meeting transcript

Meeting: Kos Team meeting (biweekly), held October 5, 2026 and organized by Kos Srivastava. Content is summarized rather than quoted at length. The learning-loop architecture is explicitly marked as an inference.

Open meeting details arrow