Verdigris Design System

Experimental

Deltas-only against pilot-kickoff. Graduates with the cell after 2 customer-facing decks ship.

Internal Team Deck

This genre is for Verdigris-internal coordination: pre-meeting prep, post-mortem reviews, weekly engagement updates, account reviews, board prep that the board hasn’t seen yet. Audience is Verdigris-only — pilot working group, leadership team, eng/GTM coordination.

This guide documents only the deltas against pilot-kickoff.md. Read that spec first.

What changes

Axis Pilot kickoff (parent) Internal team (this genre)
Length 12-20 slides 8-15
Voice primary Mike (field credibility) Thomas (operational, transparent)
Voice supporting Thomas Mike + Jon (technical translation + bench-diagnostic credibility)
Voice accent Jon (data flow / hardware) Jimit (market context when engagement intersects strategy)
CTA pattern “Pilot scope + decision date for expansion” “Decision needed by [date]” — a specific Verdigris decision, owner, date
Confidentiality default CUSTOMER-CONFIDENTIAL INTERNAL ONLY (red marking)
Date format absolute calendar dates week-N notation acceptable when the engagement timeline is the running thread
Logomark full lockup (footer + title) wordmark only (no full lockup; signals “internal context, not customer-facing”)
Default density dual-use dual-use (same as parent; some weekly reads happen async, some happen live)

The wordmark distinction matters: a Verdigris employee opening an internal deck immediately registers “this isn’t going to a customer” via the lockup variant. The customer-facing decks always use the full lockup; the internal-only decks always use the wordmark. The visual delta is small and intentional.

The density default stays at dual-use because internal team decks split delivery: weekly status decks are often read async by leads who couldn’t make the meeting; post-mortem and board-prep decks are presented live. Producers should override on the produced deck when the delivery context is unambiguous (a board-prep meeting that will not be re-shared async → density: live-spoken; a weekly status that’s mostly async-read → density: leave-behind). See pilot-kickoff.md § Presentation density for the canonical spec.

What stays the same

Skipped slides

The pilot-kickoff structure has 18 canonical slides. Internal team decks typically use a subset:

From pilot kickoff Internal team status
1. Title required
2. Why we’re here required (terse: “What this meeting is for”)
3. What we agreed to optional (covered in the linked customer-facing artifact)
4. Success criteria required (operational; tracked weekly)
5. Verdigris team required (often the title slide carries the team list inline)
6. Customer team optional
7. Timeline required (week-N notation OK)
8. Hardware install optional (when relevant)
9. Data flow optional
10. Week-4 checkpoint required
11. Risks + mitigations required (more direct than customer-facing)
12. Decisions we owe required
13. Decisions from customer optional
14. Contact + escalation optional
15. Anchor metric required
16. Path to expansion optional
17. Appendix: assumptions optional
18. Close: decision date required

Decision framework: where to land in the bounds

Boundary Floor (small) Default Ceiling (large)
Slide count (8-15) 8 slides for routine weekly status: title, why-we-here, success criteria, week-N checkpoint, risks, decisions we owe, anchor metric, close. 10-12 slides for a typical pre-meeting prep deck or mid-engagement review where extra context (data flow status, hardware install diagnostics) earns its slide. 15 slides for post-mortems or board-prep where multiple Verdigris functions (eng, GTM, finance) need their own status surface in the same deck. Above 15, the audience is no longer a single internal team and the deck should be split.

Internal decks bias toward the floor. The audience already shares context; padding the deck wastes their time. If the same internal deck recurs weekly, the producer should be cutting slides over time, not adding them.

Voice at a glance

Internal-team decks are heterogeneous: engineers need technical precision, operators need empathy, GTM needs market context. A single-voice recipe brittles. The internal_team_deck recipe runs four voices, each carrying a slice. Pulled directly from the recipe and the linked profile YAMLs.

Thomas — primary (Profile: voice/team/thomas-chung.yaml). Operational, transparent. Self-honesty is 10 in his profile, which makes him the right voice for “what’s slipping.” His thinking-out-loud register works for internal context where translation costs more than it gains.

“people should walk away feeling good”

Carries: title, why-we-here, success criteria, decisions we owe, week-4 checkpoint, close. The structural backbone.

Mike — supporting (Profile: voice/team/mike-mahedy.yaml). Translates engineering reality into operator-readable status. Field credibility grounds the customer-side perspective even when the audience is internal.

“please sanity check”

Carries: customer-side status, hardware-install status when relevant, risks + mitigations.

Jon — co-supporting (Profile: voice/team/jon-chu.yaml). Engineering status, bench-diagnostic credibility. Technical_precision is 9 in his profile and is load-bearing for engineering status slides. Internal coordination cannot lead with operations-speak alone.

“looks like a firewall issue to me”

Carries: data-flow status, hardware-install diagnostics, scope-and-coordination slides. Added 2026-05-02 after Loop 3 review caught the gap.

Jimit — accent (Profile: voice/team/jimit-shah.yaml). Connects the engagement to market signals when relevant.

“Hot new update from Semianalysis…”

Carries: occasional market-context slides when the engagement intersects strategy. Stays out of routine status decks.

Template vs. produced

Internal decks are heterogeneous and often name people by design (a weekly engagement review will say “Mike will follow up”). The template-vs-produced contract still holds for the template artifact in categories/slides/examples/; one-off produced decks fill names alongside roles.

Slot Template stage Produced stage
Title slide Weekly engagement review: <span class="vd-template">[FIELD: customer + week, e.g. "Acme Life Sciences — Week 3"]</span> Weekly engagement review: Acme Life Sciences — Week 3
Decisions we owe Pilot Lead: <span class="vd-template">[FIELD: name, e.g. "Mike Mahedy"]</span> to confirm install schedule by <span class="vd-template">[FIELD: absolute date]</span> Pilot Lead: Mike Mahedy to confirm install schedule by 2026-06-08
Risks slide Risk: <span class="vd-template">[FIELD: one-line risk]</span> — Mitigation: <span class="vd-template">[FIELD: one-line mitigation owned by named individual]</span> Risk: customer firewall blocks data egress — Mitigation: Jon Chu to coordinate with customer IT by 2026-06-15

The template stage is what an agent generates from the spec; the produced stage is what a human (or evidence-grounded agent) fills in. Never ship the produced stage without source evidence for every filled placeholder — internal does not relax this discipline, even though the audience is Verdigris-only. A “Mike will do X by 2026-06-08” claim must trace to a real Mike commitment.

Diction adjustments specific to internal decks

The diction rules for external genres (Z2O-1321, e.g., “exit criteria” → “expansion criteria”) do not apply to internal-team decks. Internal jargon is fine — and faster — when the audience is Verdigris-only.

What still applies:

Why this is a genre and not a flag on pilot kickoff

A reasonable challenge: if internal team is “deltas-only against pilot kickoff,” is it really a genre, or is it a modes: [internal] toggle (the YAML field) that could apply to any genre?

Verdict: it’s a genre, defensibly, on five structural grounds:

  1. Voice mix changes from three voices to a different three voices, with Thomas as primary instead of Mike. The voice center of gravity shifts because the audience is internal — operational clarity beats field credibility.
  2. Logomark variant changes (full lockup → wordmark only). This signals “internal context, not customer-facing” pre-consciously to readers; a wordmark-only customer deck would feel cheap, a full-lockup internal deck feels misplaced.
  3. Confidentiality tier marking color changes to red (INTERNAL ONLY), distinct from the customer/partner/public yellow/purple/grey palette.
  4. Date format relaxes to permit week-N notation when the engagement timeline is the running thread (per the absolute-dates rule’s modes: list excluding internal_team).
  5. Diction rules NOT applied — the audience-fit-diction guidance (“exit criteria” → “expansion criteria”, etc.) doesn’t apply because internal jargon is faster than translation when the audience is Verdigris-only.

If we collapsed internal_team to a modes: [internal] toggle on pilot_kickoff, every rule that distinguishes the genres would need conditional logic on the toggle. That’s complexity without simplification. Treating internal_team as a genre keeps each genre’s spec self-contained and locally readable.

The genre stays a genre. If a sixth structural delta accumulates against another existing genre and a new “post-mortem” or “retrospective” sub-genre earns its own spec, this defense is what to point at — same five-axis test.

Origin

Filed implicitly via the same pilot kickoff review (Z2O-1318 through Z2O-1323) — the slide cluster surfaced in customer-facing decks but the rules need to NOT apply to internal decks (e.g., the customer-101 wordmark would be wrong for an internal weekly), so the genre split was necessary to express the deltas. The four-genre framework was synthesized via the adversarial-review workflow.

What this cell does NOT cover

See also