Run a quarterly initiative audit from scattered notes

Why this matters

Roadmaps love starts. Operating reality is threads, half-written docs, and Slack shards nobody reads end-to-end before the quarterly session. Without a forced portfolio view, teams keep recycling the same polite ambitions. Every initiative stays “active” until the budget argues back.

An initiative audit is not a status report. It answers four blunt questions once per quarter: what we are actually pushing, what is coasting on slogans, what disappeared without a funeral, and what we will stop pretending to fund. That last bucket is the scarcest artifact in most planning decks.

Today you mint a partner that turns pasted sources into those buckets. It stays honest because you supplied the paste, not because the agent inferred your roadmap from silence.

The next time leadership asks what matters this quarter, you’ll have one shared table and a not doing list that nobody has to invent live on the call.

Quick start

1. Open objective guardrails: paste your approved initiative names and definitions from a planning doc or context file (Add your first context file). Forbid inventing new portfolio labels mid-audit.
2. Create Initiative Audit: ask the Admin Agent in Chat for an agent whose charter demands three buckets (Pursued, Thin, Absent) plus a NOT_THIS_QUARTER section. Each row cites SOURCES_QUOTED from what you pasted or says UNKNOWN.
3. Gather your quarterly sources: combine three inputs minimum: a weekly rollup snippet (Roll up initiative threads into weekly goal check-ins), one leadership or planning paragraph, and bullets from Turn meeting notes into action items and decisions. Redact names to roles where needed.
4. Run the audit: paste the packet. Reject the first answer if any row lacks a receipt sentence tied to quoted text. Tighten once.
5. File the artifact: save markdown with the quarter label in the filename. An optional reminder (Run a workflow) only nags you to gather sources. Auxot does not spider Slack unattended.

Done? One completed audit table where Thin rows hurt to read. That is proof the buckets are doing work.

The agent can do that?

You classified the portfolio once. These three prompts keep the ritual from turning into an empty morale exercise.

1. Definitions before debate

Chat → Admin Agent:

Paste objectives excerpt only: […]. Write Pursued / Thin / Absent definitions for OUR vocabulary (<=8 lines each). Pursued means staffed cadence and decisions this quarter. Thin means talk without owners or dates. Absent means not observable in sources. Forbid numeric fake progress. Markdown. No rows about specific initiatives yet.

Why it’s non-obvious: Without frozen definitions, Thin becomes a polite synonym for later. Executives cannot compare quarter four to quarter one until the buckets stay stable.

2. Stop-pretending paragraph

Audit table pasted: […]. Draft **STOP_PRETENDING** (max five sentences). Name initiatives we should quit claiming. Each sentence ties to a Thin or falsely Pursued row. If evidence is missing, say **UNKNOWN** instead of drama. End with who must own the awkward conversation.

Why it’s non-obvious: Kill lists usually rest on gut feel, not evidence. Forcing receipts surfaces organizational debt you can put on the calendar instead of discovering in the hallway.

3. Inverted brief (argue for fewer rows)

Assume we must drop half the Pursued rows without adding headcount. Pick survivors (max four). Rewrite Thin rows as Absent or NOT_THIS_QUARTER (one sentence each). Brutal tone. No new initiatives invented for symmetry.

Why it’s non-obvious: Planning bias adds work. Inversion proves which initiatives still look worth funding before the calendar invite goes out.

Go deeper

Weekly vs quarterly

Weekly roll-ups compress seven days (Roll up initiative threads into weekly goal check-ins). This audit asks whether the initiative still deserves real calendar time. That is a complementary rhythm, not a duplicate spreadsheet.

AI fleet vs business portfolio

Fleet reviews track agents and automation (Run a quarterly review of your agents). Initiative audits track the initiatives your humans swore mattered. Bring both documents under separate headings so budgets do not get conflated by accident.

Longer horizons

Narrative sequencing (Plan your first 90 days with your agents): audit output feeds your quarterly planning checkpoints. When you reuse this language in other docs, keep the same definitions; don’t let the bucket names drift.

Framing for leadership

For exec polish, lean on Brief leadership on your agent program: your STOP_PRETENDING paragraphs should borrow that briefing tone.

Walkthrough

Step 1: Cap initiatives reviewed

A deep audit of a few initiatives beats a shallow pass across twenty-eight symbolic rows. Pick the eight your exec team names aloud. Refuse the urge to expand the list.

Step 2: Freeze receipts

SOURCES_QUOTED must cite pasted fragments, not links you did not include. Tighten the output until made-up citations disappear.

Step 3: Pair with calendar reality

If nobody schedules initiative reviews between audits, Thin repeats. Add a human checkpoint (Require human approval before risky actions) only where the dollars or reputational risk are real. Skip bureaucracy elsewhere.

Step 4: Archive quarterly

Store markdown beside finance slides so it stays searchable, not only Chat scrollback. Compare Absent rows quarter-over-quarter for silent deaths.

Step 5: Separate fleet time

Scheduling this in the same afternoon as the AI fleet review tempts you to blend agendas. Run two headings: humans before models. That keeps both agendas honest.

What’s next

• → Roll up initiative threads into weekly goal check-ins. Weekly receipts feed quarterly classification. Paste rollup excerpts into the audit packet.
• → Run a vendor and stack creep inventory. Roadmap honesty includes SaaS sprawl and unknown owners; same explicit not doing tone.
• → Run a quarterly review of your agents. Automation portfolio, same calendar discipline, different scorecard.
• → Turn meeting notes into action items and decisions. Collect initiative mentions before threads evaporate. Paste synthesis rows into SOURCES_QUOTED.
• → Brief leadership on your agent program. STOP_PRETENDING tone alignment and exec-facing polish.
• → Build a cross-meeting commitments ledger. Promises across meetings. Initiative audits ask whether the promise still matches staffing (adjacent lens).

Reference

• Pages in Auxot: Chat, Settings → Agents, Settings → Context Files, and Workflows
• See also: Wire your first tool with least privilege, Plan your first 90 days with your agents, Track strategic decisions and when to revisit them, Run a vendor and stack creep inventory, Audit and clean up your agents, Build your first Dashboard