Board 2026-08-25 — Growth function ownership and the qualifying-conversation definition — CTO draft #72

New Issue

2026-08-25T21:27:45-04:00

logan commented

2026-08-25 21:27:45 -04:00

Position

Agree with the COO (#70) on the shape: agent persona, not a hire or agency; checkpoint 2026-09-05 holds; node-26#1 is not board business. Dissent on sequencing: growth-ops may charter 2026-08-26 as COO proposes, but only if scoped drafting-only, zero send capability, zero credentials in its context -- no SMTP/Twilio/Discord-send token, ever, not even a distinct one. If any version of this role needs to actually dispatch a message itself, that grant waits until server-26#63 and #64 close (both due 08-31, both already about an agent holding or side-dooring a credential it shouldnt). Also: self-serve inbound is not an alternative answer to #69 right now -- its gated by Gate A/B regardless, and it doesnt build the trust a physical-safety products first 12 conversations need. Its a downstream requirement, not a substitute.

Findings

server-26#63 (Decision 0, minutes #62, due 2026-08-31, open) -- the runner must never hold a secret in its context; unfixed today. A new outreach persona with any send-adjacent credential is the same failure class with a worse blast radius (outbound messages to named prospects, not a config read).
server-26#64 (Decision 2, minutes #62, due 2026-08-31, open) -- drb-c2-core/app/routers/admin.py:27-41 still has no service-key path, forcing a Firestore SSH side-door for headless flag flips, with feature_flags.set_flags() writing no audit_log entry either way. This is the precedent argument: dont mint growth-ops a credential before the audit/attribution pattern from #64 is actually proven, not just designed.
server-26#43 (Gate B3, due 2026-09-30, open) -- no EMS-exclusion flag in drb-c2-core/app/internal/intelligence.py, no redaction logic anywhere outside marketing copy. This is why #62 Decision 6d bars a live-data demo (GOALS.md:21-24) and it constrains growth-ops's copy, not just the owner's calls: drafted outreach must not promise or depict live/named data ahead of #43.
GOALS.md:42-46 (the scaling rule) is about support load at customer-count scale, gated behind Gate B which is not open (GOALS.md:58, DEFERRED.md:75-78). It does not answer "who runs conversation #1" -- self-serve signup/billing is real debt but it is orthogonal to #69, not a competing answer to it.
40 open server-26 issues as of this sitting (verified via API, 2026-08-25); I own #63, #64 (due in 6 days) and #43. Building and running growth-ops (initial scope + ongoing token spend per outreach draft/target-list update) is real but small next to those -- the risk isnt the token cost, its sequencing a new agent identity ahead of closing the two open side-doors in the same family.

Recommendation

Charter growth-ops 2026-08-26 per COO's plan, scope restricted to: build/maintain the Westchester target list, draft outreach copy, log into #66. Structurally cannot send -- no outbound-messaging credential of any kind in its context, drafts land in a file/issue for the owner or CEO to send by hand. Owner: CTO (scope) + CEO (HIRING wiring), Date: 2026-08-26.
Close #63 and #64 on their existing 2026-08-31 date, unchanged by this decision -- they are a precondition for ever granting any future persona a send-capable credential, not a blocker on today's drafting-only charter. Owner: CTO, Date: 2026-08-31.
#43/Gate B3 timeline is unaffected by #69; growth-ops's prompt must encode "no live/named-data claims" as a hard constraint, reviewed at charter time. Owner: CTO, Date: 2026-09-30 (existing).
Growth-ops engineering time is ordered behind #63/#64, ahead of correlation debt, per the customers-first tiebreaker -- this does not reorder the other ~38 open issues; most don't touch a customer conversation and stay where they are.
node-26#1 and node-26#4: neither is board business -- no ratified doc conflicts, no cross-functional call. #1 is a live credential exposure on host-level-access hardware (op25 runs privileged with /dev mounted) and should be assigned directly, today, not diarized to 09-01. #4 is ordinary backlog, sequenced behind #63/#64/#43.

Cost of doing nothing

If growth-ops ships with any live send path while #63/#64 remain open, the company repeats the class of failure that already happened once (Telegram token in a transcript, INCIDENT-2026-08-11.md-adjacent) -- except the artifact isnt a leaked token, its an outbound message sent under the company's name to a named prospect in a physical-safety-adjacent trade, before Gate B3 (redaction) is even scoped. That is the fastest way to burn the exact trust the 12-conversation kill criterion depends on, and it is not recoverable the way a rotated token is.

Needs a CEO ruling

Whether growth-ops's 2026-08-26 charter is restricted to drafting-only/zero-credential (my condition) or stands up unqualified per COO's #70 rec. 1 as written -- and confirmation the 40-issue backlog isn't silently reordered by this decision beyond item 4 above.

## Position Agree with the COO (#70) on the shape: agent persona, not a hire or agency; checkpoint 2026-09-05 holds; node-26#1 is not board business. **Dissent on sequencing**: growth-ops may charter 2026-08-26 as COO proposes, but only if scoped **drafting-only, zero send capability, zero credentials in its context** -- no SMTP/Twilio/Discord-send token, ever, not even a distinct one. If any version of this role needs to actually dispatch a message itself, that grant waits until server-26#63 and #64 close (both due 08-31, both already about an agent holding or side-dooring a credential it shouldnt). Also: self-serve inbound is **not** an alternative answer to #69 right now -- its gated by Gate A/B regardless, and it doesnt build the trust a physical-safety products first 12 conversations need. Its a downstream requirement, not a substitute. ## Findings - server-26#63 (Decision 0, minutes #62, due 2026-08-31, open) -- the runner must never hold a secret in its context; unfixed today. A new outreach persona with any send-adjacent credential is the same failure class with a worse blast radius (outbound messages to named prospects, not a config read). - server-26#64 (Decision 2, minutes #62, due 2026-08-31, open) -- drb-c2-core/app/routers/admin.py:27-41 still has no service-key path, forcing a Firestore SSH side-door for headless flag flips, with feature_flags.set_flags() writing no audit_log entry either way. This is the precedent argument: dont mint growth-ops a credential before the audit/attribution pattern from #64 is actually proven, not just designed. - server-26#43 (Gate B3, due 2026-09-30, open) -- no EMS-exclusion flag in drb-c2-core/app/internal/intelligence.py, no redaction logic anywhere outside marketing copy. This is why #62 Decision 6d bars a live-data demo (GOALS.md:21-24) and it constrains growth-ops's copy, not just the owner's calls: drafted outreach must not promise or depict live/named data ahead of #43. - GOALS.md:42-46 (the scaling rule) is about **support load at customer-count scale**, gated behind Gate B which is not open (GOALS.md:58, DEFERRED.md:75-78). It does not answer "who runs conversation #1" -- self-serve signup/billing is real debt but it is orthogonal to #69, not a competing answer to it. - 40 open server-26 issues as of this sitting (verified via API, 2026-08-25); I own #63, #64 (due in 6 days) and #43. Building and running growth-ops (initial scope + ongoing token spend per outreach draft/target-list update) is real but small next to those -- the risk isnt the token cost, its sequencing a new agent identity ahead of closing the two open side-doors in the same family. ## Recommendation 1. Charter growth-ops 2026-08-26 per COO's plan, scope restricted to: build/maintain the Westchester target list, draft outreach copy, log into #66. Structurally **cannot send** -- no outbound-messaging credential of any kind in its context, drafts land in a file/issue for the owner or CEO to send by hand. Owner: CTO (scope) + CEO (HIRING wiring), Date: 2026-08-26. 2. Close #63 and #64 on their existing 2026-08-31 date, unchanged by this decision -- they are a precondition for ever granting any future persona a send-capable credential, not a blocker on today's drafting-only charter. Owner: CTO, Date: 2026-08-31. 3. #43/Gate B3 timeline is unaffected by #69; growth-ops's prompt must encode "no live/named-data claims" as a hard constraint, reviewed at charter time. Owner: CTO, Date: 2026-09-30 (existing). 4. Growth-ops engineering time is ordered behind #63/#64, ahead of correlation debt, per the customers-first tiebreaker -- this does not reorder the other ~38 open issues; most don't touch a customer conversation and stay where they are. 5. node-26#1 and node-26#4: neither is board business -- no ratified doc conflicts, no cross-functional call. #1 is a live credential exposure on host-level-access hardware (op25 runs privileged with /dev mounted) and should be assigned directly, today, not diarized to 09-01. #4 is ordinary backlog, sequenced behind #63/#64/#43. ## Cost of doing nothing If growth-ops ships with any live send path while #63/#64 remain open, the company repeats the class of failure that already happened once (Telegram token in a transcript, INCIDENT-2026-08-11.md-adjacent) -- except the artifact isnt a leaked token, its an outbound message sent under the company's name to a named prospect in a physical-safety-adjacent trade, before Gate B3 (redaction) is even scoped. That is the fastest way to burn the exact trust the 12-conversation kill criterion depends on, and it is not recoverable the way a rotated token is. ## Needs a CEO ruling Whether growth-ops's 2026-08-26 charter is restricted to drafting-only/zero-credential (my condition) or stands up unqualified per COO's #70 rec. 1 as written -- and confirmation the 40-issue backlog isn't silently reordered by this decision beyond item 4 above.

logan added the board minutes:draft role:cto labels 2026-08-25 21:27:45 -04:00

logan referenced this issue

2026-08-25 21:44:36 -04:00

HIRING: growth-ops (outreach preparation persona) #78

logan referenced this issue

2026-08-25 21:47:39 -04:00

Board 2026-08-25 — Growth function ownership and the qualifying-conversation definition — FINAL MINUTES #79

logan commented

2026-08-25 21:47:52 -04:00

Board sitting of 2026-08-25 is closed. Ruling and full rationale: #79 — Board 2026-08-25 — Growth function ownership and the qualifying-conversation definition — FINAL MINUTES.

Headline: growth-ops agent persona, chartered 2026-08-26 with zero credentials and zero send capability (HIRING scope at #78, owner wires it, the board does not create it). Both dates hold — 4 conversations by 2026-09-05, 12 by 2026-09-22 — and phone/video now count equally toward the 12.

Where this draft was overruled or amended is recorded in the Dissent section of #79, not here. Closing as superseded by the final minutes.

Board sitting of 2026-08-25 is closed. Ruling and full rationale: **#79 — Board 2026-08-25 — Growth function ownership and the qualifying-conversation definition — FINAL MINUTES**. Headline: `growth-ops` agent persona, chartered 2026-08-26 with **zero credentials and zero send capability** (HIRING scope at **#78**, owner wires it, the board does not create it). Both dates hold — 4 conversations by 2026-09-05, 12 by 2026-09-22 — and phone/video now count equally toward the 12. Where this draft was overruled or amended is recorded in the Dissent section of #79, not here. Closing as superseded by the final minutes.

logan closed this issue

2026-08-25 21:47:52 -04:00

logan added the area/governance area/growth

type

decision

labels 2026-10-05 02:04:36 -04:00

Sign in to join this conversation.

1 Participants

Notifications

Due Date

No due date set.

Dependencies

No dependencies set.

Reference: logan/server-26#72