role
Sam — Brain Content Lead JD
Sam's Brain Content Lead role definition: authors the curated global brain, reviews cross-pollinated entries, operationalizes the voice/style guide, runs the content-side brain audits, and enforces sensitive-vertical exclusions. Reports to Santiago Perez Asis.
Role: Brain Content Lead, MemberIntel
Incumbent: Sam (internal promotion from the MemberPress support team, approved at the 2026-05-11 Blair × Seth working session in lieu of an external hire; last name TBC for publication)
Reports to: Santiago Perez Asis (Product + Project Lead, MemberIntel)
Peers: Russ Williams (Lead Designer — co-owns voice / style guide), Senior AI Engineer (owns the technical eval suite; you own the content-quality counterpart)
Effective: TBC — contingent on the MemberPress support-assist light reducing Sam’s ticket load. Date not yet committed.
References: V1 spec §6 (the brain), V1 spec §6.3 (cross-pollination), V2 spec §6 (platform-tagged brain), arch-cross-pollination, Santiago JD, Russ JD, Senior AI Engineer JD, privacy strategy for counsel
Mission
Own the content quality of MemberIntel’s brain — the curated playbook that gives the AI its MemberPress-operator expertise. Author and curate the global brain. Review per-customer brain entries when they’re promoted via cross-pollination. Be the human in the loop who makes sure the AI sounds like a real MP operator and says useful, accurate, well-bounded things.
The brain is the moat (V1 spec §6.4). It is also the deepest reason a customer trusts (or doesn’t trust) the AI. This role is load-bearing.
This is a content + subject-matter-expertise role, not an engineering role. The Senior AI Engineer owns the technical eval suite — prompt regression, citation accuracy, tier-routing safety, hallucination rate. You own the content evals: does this entry sound right, is the playbook actually useful, is the voice consistent, would a real MP operator nod or wince. The two halves connect through monthly audit reviews; neither half ships without the other.
Authority structure
Blair (CEO, MemberIntel) holds:
- Product strategy, target customer, 18-month roadmap
- Final approval on PRDs and material brand-voice direction
Santiago Perez Asis (Product + Project Lead) holds:
- PRD authoring, sprint scope, ship-gate calls
- Voice / style guide direction (co-owned with Russ; you operationalize)
- Cross-functional coordination outside the brain content team
Russ Williams (Lead Designer) holds:
- Voice / style guide creative direction (co-owned with Santiago)
- UI surfaces where brain content renders — chat answer rendering, citation styling, weekly digest layout
Senior AI Engineer (peer) holds:
- The technical eval suite (regression, citation accuracy, tier-routing safety, hallucination rate)
- Prompt engineering and prompt versioning
- The cross-pollination job’s execution plumbing — you own its content judgment; they own its anonymization and k-anonymity floor
Brain Content Lead (this role) holds:
- Authoring the curated global brain (target: 200+ approved entries by month 6 per V1 spec §2)
- Reviewing every cross-pollinated entry before it lands in the global brain
- Voice / style consistency across all brain content (operationalize the guide Russ + Santiago co-deliver)
- Periodic brain content audits — the content counterpart to the AI Engineer’s regression suite
- Sensitive-vertical exclusion enforcement (medical, advocacy, etc.) — human judgment above the k-anonymity floor
- Synthesizing thumbs-down feedback into new brain entries
- (V2+) BuddyBoss-tagged brain playbooks
What you own
1. Global-brain authoring (per V1 spec §6.1)
The curated playbooks that ship in V1 and grow with the product. MemberPress-operator how-tos, pricing guidance, content-protection patterns, dunning playbooks, member-engagement plays, churn-recovery tactics, onboarding sequences. The kind of advice an experienced MP operator would give another MP operator — except now it’s in the AI’s hands.
You write it. Target: 200+ approved global-brain entries by month 6 of V1 launch (per V1 spec §2 success metric #4).
2. Cross-pollination review (per V1 spec §6.3 and arch-cross-pollination)
Every entry promoted from a per-customer brain into the global brain via cross-pollination goes through you. The pipeline enforces the k-anonymity floor and the anonymization step (Senior AI Engineer owns that surface). Your job is the human judgment above the floor:
- Is this actually generalizable beyond the source customer?
- Is it accurate?
- Is it well-said in MemberIntel’s voice?
- Does it belong in the global brain, or only in the source customer’s brain?
- Has identity been fully stripped — not just the obvious fields, but the case-specific tells a human reviewer can catch?
Review SLO: 95% of cross-pollinated entries reviewed within 7 days of staging.
3. Voice / style guide (operationalize)
Russ + Santiago co-deliver the guide. You operationalize it — every brain entry is reviewed against it. When the guide doesn’t cover a case, you escalate to Russ / Santiago for an update, and you carry the new rule forward.
The voice rule isn’t decorative. A customer who notices three different voices in a week stops trusting the AI. Voice consistency is a measurable, defensible quality of the brain.
4. Brain content audits
Monthly reviews of a sampled cross-section of brain content for:
- Accuracy. Has anything aged out? Has MemberPress shipped a feature that makes this entry wrong?
- Voice. Does it still sound like one operator?
- Usefulness. Is the AI citing this entry a lot but the customer thumbs-downing the answer? That’s a content problem, not a prompt problem.
Surface findings to Santiago + the Senior AI Engineer. This is the content counterpart to the AI Engineer’s regression suite — both halves run monthly and the read-outs are paired.
5. Sensitive-vertical exclusion enforcement (per privacy strategy)
Per the privacy strategy: medical, advocacy, and other sensitive verticals are permanently excluded from the global brain. The pipeline blocks the obvious cases. You are the human who catches the non-obvious ones.
Zero leaks is the target. If you’re unsure, it doesn’t go in. The cost of one bad cross-pollination is higher than the cost of ten conservative rejections.
6. Brain growth from real customer feedback
The Senior AI Engineer’s pipeline surfaces thumbs-down trends as eval gaps. Many of those gaps are content gaps — the brain didn’t know the answer. You take those signals and author the missing entries.
Turnaround target: median ≤14 days from signal to published entry.
7. V2 BuddyBoss-tagged content (per V2 spec §6)
V2 introduces platform-tagging in the brain. You author the BuddyBoss-flavored playbooks — community engagement, paid groups, member-to-member networking, BB-app integration patterns.
Target: ≥30 BB-tagged playbooks at V2 launch, growing to ≥60 within 6 months (per V2 spec §2 success metric #4).
What you do NOT own
- The technical eval suite — Senior AI Engineer
- Prompt engineering and prompt versioning — Senior AI Engineer
- The cross-pollination job’s plumbing — k-anonymity floor, anonymization step, scheduling — Senior AI Engineer + Ronald
- Product PRDs, ship-gate calls — Santiago
- UI / design — Russ
- Brand voice strategy at the marketing level — Blair (you operationalize at the brain level; marketing voice is a separate surface)
- Continuing MP support tickets at the prior scope — the support team continues with the support-assist light; you transition out gradually as that ships
- The Pro vs Free differentiation in chat quality — Senior AI Engineer (you produce the content; the model and prompt make it sing or stumble per tier)
If asked to weigh in outside this list, route to the right owner.
Critical role norms
- Voice consistency is non-negotiable. A customer who notices three different voices in a week stops trusting the AI. The voice guide is the floor; your judgment is above it.
- Sensitive verticals are a hard exclusion. Not “we’ll be careful” — actually zero entries. If you’re unsure, it doesn’t go in. The conservative call is the right call.
- Cite the source customer’s pattern, never the source customer. Cross-pollination strips identity; your review reinforces that strip. If a reviewer at any future point can re-identify the source from the entry, the entry is wrong.
- The brain is for real operators. Generic AI-shaped advice doesn’t ship. If a real MP operator would read an entry and say “no kidding,” it’s not useful — rewrite or kill it.
- Feedback is content debt, not someone else’s problem. When the AI Engineer surfaces a thumbs-down trend, the next entry on your queue is the gap that caused it.
- No personal customer data in the global brain. Ever. Patterns, yes. Specifics, no. The privacy commitment to customers depends on this being airtight.
- Disagreements with Santiago go to Blair within 48 hours. Same escalation discipline as the rest of the team.
Success measures (12-month)
| Measure | Target |
|---|---|
| Transition fully onto MemberIntel from MP support | Date TBC — when MP support-assist light has reduced ticket load |
| Approved global-brain entries by month 6 of V1 launch | 200+ (per V1 spec §2) |
| Cross-pollinated entry review SLO | 95% reviewed within 7 days of staging |
| Voice consistency in monthly brain audits | ≥90% of sampled entries pass on first review |
| Sensitive-vertical leaks into global brain | 0 incidents |
| BB-tagged brain entries at V2 launch | ≥30; ≥60 within 6 months of V2 launch (per V2 spec §2) |
| Thumbs-down → new-entry turnaround | Median ≤14 days from signal to published entry |
| Coordination with the Senior AI Engineer’s regression read-outs | Paired monthly review, both halves green |
Reporting cadence
| Cadence | Audience | Format |
|---|---|---|
| Weekly | Santiago | 30-min content sync — backlog, blockers, brain growth |
| Bi-weekly | Russ + Santiago | Voice / style guide working session |
| Monthly | Santiago + Senior AI Engineer | Brain audit review — content quality + technical eval correlation |
| Monthly (from Phase 4) | Blair | Brain growth + thumbs-feedback closure rate (Santiago presents; you provide the data) |
Dependencies and open questions
- Transition timing. Sam can’t fully ramp until the MP support-assist light has reduced his support ticket load. Date not yet committed — Blair to confirm.
- Possible second hire. A second technically-inclined support agent may be added to the brain content team, scaled to need. Decision deferred.
- Voice / style guide owed. Russ + Santiago co-deliver the guide before Sam can fully ramp. Currently scoped, not yet in flight.
- Smart distillation model (V1.5+). A paid model may be used to draft brain entries from raw customer data even on the Free tier. If shipped, Sam reviews and edits the drafts rather than authoring from scratch — capacity multiplier, not a substitute for human judgment.
What “good” looks like in this role
- A customer reads a brain-cited answer and thinks “yes, that’s what I’d tell another MP operator.”
- Voice is consistent across the whole brain. New entries from cross-pollination feel like they belong.
- Sensitive-vertical exclusions hold. Zero incidents.
- When a thumbs-down trend appears, the gap closes in days, not weeks.
- V1 ships with a brain that’s already useful — not a stub waiting for content.
- V2 ships with a BuddyBoss-flavored brain that BB customers immediately recognize as theirs.
- When Blair asks “is the brain working?”, the Senior AI Engineer answers with eval numbers and you answer with audit results — and both are green.
Document version: Draft v1 — to be reviewed with Sam before finalization. Decision-rights surfaces require sign-off from Blair, Santiago, and Sam before this JD goes operational. Open questions: transition date (gated on the MP support-assist light), whether to expand to a second brain content reviewer, and final title (Brain Content Lead is preferred over “Brain Eval Engineer” — this is a content role with quality-review responsibilities, not an engineering role).