Concrete interventions, ordered by leverage. Each opens to a specific behavior, an owner, and a way to know it has stuck. Filter by tier; click any item to expand.
The single highest-leverage intervention. Sandy's structural corrections land 60–70% of the time today; with explicit timekeeper / agenda-enforcement authority, the rate becomes 90%+.
Greg has the analytical instinct for agenda design but Sandy has the willingness to act on the discipline in real time — propose Sandy as primary co-facilitator with Greg as the meta-auditor in the close.
Scorecards have been skipped in every single meeting because no metrics exist. Sandy's KPI rock has been deferred or softened multiple times. The team cannot run an L10 without a scorecard segment.
Propose a 5-metric "minimum viable scorecard" live by 2026-05-29:
Imperfect metrics > no metrics.
Adopted in principle (4/22) but the practice has reverted. Make it the default:
Total 70 minutes — leave 20 minutes of buffer for the demo / tangent that will inevitably happen.
Brayden's tool-builds (kernel system, Obsidian dashboards, Aiyarise app) consume 15–35 minutes when they happen. Create a separate weekly 30-minute "Show & Tell" optional slot (e.g., Friday 11 AM) where artifacts get demoed without competing with the L10's decision-time. Use Loom for asynchronous demos when possible.
No issue is opened until every attendee has had a named opportunity to weigh in. When an IDS issue surfaces, the chair names every team member and gets a one-sentence position before the discussion expands. Eliminates the Jeremy-disappears and Victoria-withdraws patterns.
Greg implicitly requested this on 4/22. 10 minutes per issue. Three issues maximum. If unresolved at the timer, schedule a follow-up rather than extending in-meeting. The Kasian IDS would have been 30 minutes instead of 45 with this discipline on 5/6.
Brayden has joked about it repeatedly; Greg has flagged it gently twice. Propose a peer-care norm: any time Brayden discloses an all-nighter, the chair responds with one of three structured prompts:
Light, not punitive. The norm exists; the discipline doesn't.
Jeremy is now the only AI Systems Developer. Greg's May 29 deliverable is at risk. The team's technical capacity has materially contracted. Spend 30 minutes (off-L10) re-allocating Q2 rocks against current capacity. Decide explicitly: hold the rocks, push the dates, or reassign owners. Do not let the original allocation drift silently.
Many of the patterns that fail to land in the L10 (tooling discipline, scope creep, mission-statement drift, ICP discipline) are best addressed in a 30-minute weekly 1:1 rather than in front of the team. Sandy already references "I'd love to set up a one-on-one" multiple times. Make it a calendar fixture.
Brayden's phone interrupted 4/15 (twice — second time he stepped out for the call). Greg's daughter has called twice. Adopt a single rule: if it can wait 30 minutes, wait. If it can't, the person handling it owes a one-line summary back to the team.
The spelling corrector should run on every transcript as a default, not on-demand. Already documented as a master-CLAUDE.md rule for downstream analysis — but the principle applies to the team's own use of transcripts (action-item extraction, follow-ups).
Multiple risks (Kasian contract, agent safety, fear-based-branding workshop, security expertise) have been raised, agreed, and dissolved without an owner. Adopt a hard rule: no IDS issue closes without (a) owner name and (b) ISO-format date.
Built 4/15. Has commercial potential Brayden flagged immediately (specifications review, tender risk analysis). Has not been productized. Spend 30 minutes scoping whether this becomes a paid offering, a lead-magnet, or a discontinued experiment.
Surfaced 4/29. Agreed to be a 3–4 hour or full-day session. Not yet scheduled. The team building an AI governance consultancy is currently shipping fear-laden marketing language. Run the workshop.
Five missed meetings is too many. Either formalize his attendance expectation (every L10) or move him to a different cadence (monthly strategic, with explicit no-L10 expectation). The current state — assumed-present, frequently absent — produces friction for Greg and erodes the team's ability to plan around his contributions.