90-day charter

Write this in week 1–2 with my manager. Keep the living copy offline if it has internal names. This page is the template + how I actually use it.

Purpose (say this to myself first)

Force AI platform ownership into the seat — agents, RAG, governance, evals — so chapter 4 panels hear production judgment from this role, not only Lucid plus open demos.

A Staff+/Principal 90-day plan is closer to a manager’s listening plan than a senior IC “land my first PR” plan. Leverage is force multiplication: standards, gates, and decision rights other teams inherit. Shipping a flashy agent in week three that nobody else can operate is a senior-engineer win and a Principal miss.

What good charters do

Listen-first (don’t skip)

Days 1–14 — diagnose before I decree

Talk to: manager, the person who owns auth/IAM, whoever owns the noisiest agent or RAG path, one skeptical senior engineer, one business partner who feels the pain. Ask:

1. What irreversible AI action would embarrass us on the front page?
2. Where does retrieval happen relative to authorization today — honestly?
3. What has already been tried and quietly abandoned?
4. Which team would adopt a standard if we wrote a thin one?

I’m not collecting friends. I’m finding the fulcrum: the place where one gateway, one eval gate, or one authZ-before-retrieve path multiplies others.

Days 15–30 — lock the charter

Propose three outcomes. Get email/doc agreement. Put 30/60/90 on the calendar.

Days 31–90 — earn the right to set direction

Small wins that build trust, then the harder standard. If I reverse a popular but unsafe autonomy default, I document the incident that justified it — that’s Principal storytelling later.

Charter one-pager (fill)

Role context

• Title / level: ________
• Manager: ________
• Team mission (1 sentence): ________
• Transition type (honest): start-up inside a mature org / realignment / sustaining

What I say to my manager (north-star alignment, not a threat)

I want the first 90 days to produce something another team can reuse — a deny path, HITL on a scary tool, or an eval that can fail a build. I’ll listen two weeks, then lock three outcomes with you. If this is mostly CRUD with AI in the title, we should say that now.

Outcomes (pick 3 — measurable)

Strong shapes:

• AuthZ-before-retrieval or tenant isolation path documented + negative evals in CI
• Gateway / HITL on one irreversible tool path (CRM write, deploy, publish, refund, payout)
• Eval regression gate on one production agent/RAG flow — build fails on quality drop
• Org ADR or reference architecture adopted by ≥1 other team
• Cost/latency budget + FinOps meter on one high-traffic agent path

Weak: learn codebase; POC chatbot; “AI strategy deck”; demo day without a production owner.

Explicit non-goals (write 3)

Examples I can steal and edit:

1. No org-wide multi-agent rewrite in Q1
2. No new public product brand from this seat
3. No replacing the existing model provider in the first 60 days unless risk demands it

Support I need

• Access / data / SSO: ________
• Pair / reviewer: ________
• Decision rights (what I can ship without a committee): ________
• What requires VP / security / legal: ________

Check-ins

• Week 2: charter signed
• Day 30 / 60 / 90: outcome review with manager — evidence, not vibes

Seat-type adaptations

External interviews still use one personal brand.

First-week script (manager 1:1)

I’m not presenting a McKinsey pack.

1. Sixty seconds on Lucid outcomes I can defend — not the catalog.
2. “Here’s what I’m afraid of if we only ship demos…” — irreversible tools, retrieval before authZ, evals that never fail builds.
3. Propose three candidate outcomes; ask which scare them most if missing.
4. Ask: What irreversible AI actions scare this org most?
5. Ask: Which team will adopt a standard if we ship a thin one?
6. Ask: What does success at day 90 look like in your words?
7. Get it in writing. Email is enough.

If they push back (“just ship features”)

I can help ship. I’m asking for a thin ownership charter so the AI work has a control plane — otherwise we’ll recreate the demo pile every quarter. Pick one irreversible path and one eval gate with me. I’ll take the rest as product delivery.

Day 30 review with manager:

Here’s what shipped: [outcome 1 evidence]. Here’s what I won’t boil: [non-goal]. Here’s the one standard I want another team to reuse next. Does that still match what success looks like in your words?

If they still refuse any AI ownership, that’s a signal for chapter 4 timing, not a personality conflict. I document it privately.

If they ask for a chatbot in week two

I can help ship a thin wedge. I’m still going to put HITL on the irreversible write and an eval on the path. A demo without those is a week-two win and a quarter-later incident.

Interview: “Walk me through your 90-day plan”

I don’t start by rewriting the platform. First two weeks I’m mapping decision rights, tool surfaces, and where retrieval sits relative to authorization — and I’m listening for the irreversible action the org is already nervous about. Then I agree three outcomes with my manager: typically harden one deny-path for tenancy or authZ-before-retrieve, put HITL or a gateway on one irreversible tool, and put an eval gate on one flow that can fail CI. Non-goals get written down so we don’t boil the ocean. By day 90 I want at least one standard another team will actually reuse. That’s the test — not a prettier chatbot.

How this feeds chapter 4 stories

Every charter win becomes a STAR later:

• S/T — org risk I found in the listen phase
• A — the gate/standard I shipped and who adopted it
• R — metric, adoption, or incident prevented — with provenance

No charter → chapter 4 panels stay stuck on Lucid + public demos. Still workable. Just thinner.

Related

• Stabilize and aim up
• Signal artifact — often becomes outcome #2 or #3
• Leveling narrative