M MemberIntel KB
⌘K

reference

Decisions Due — Monday 2026-05-11 (prep doc — historical)

PREP doc for the 2026-05-11 meeting. The actual outcomes are at /meeting-2026-05-11-blair. References to 'Product Lead' below are historical — the role was retired in this session; Santiago absorbed the duties.

Status (2026-05-11): This was the prep doc Blair was reading into. The meeting happened. For the actual decisions and action items, see /meeting-2026-05-11-blair. References to a “Product Lead” below are now historical — Cindy’s role was not backfilled in this session; Santiago absorbed the duties (see /santiago-jd).

Drafted: 2026-05-09 (evening)
For: Blair Williams, CEO
Authoritative role doc: Blair — CEO JD
Format: Read before the meeting. ~10 minutes.
Methodology: Generated by the persona dev loop — ceo-blair did the inventory triage, product-lead and ai-engineer cross-checked from their JD lenses. Persona artifacts in docs/superpowers/reviews/ (look for 2026-05-09-*-monday-prep-*). Each call below is the synthesis.

What this meeting is, in your terms

This is the bridge from “architecture is locked” to “build can start clean.” Five ADRs landed in the past week (good — Seth held the line on writing them down), but two of them want your explicit signature, and one surfaces a real economics question that’s yours, not engineering’s. Most of the 34 inline open-questions across the v1, v1.5, and v2 spec pages aren’t actually yours — they’re either tactical execution the leads should run, or they got resolved by the ADRs and just haven’t been crossed off the list. The handful that ARE yours cluster around three things: pricing/launch packaging, the Free-tier circuit-breaker design, and the Anthropic dependency posture you half-decided May 5.

What success looks like: three deliberate decisions resolved, six confirmations approved (or redirected), four status updates absorbed. Walk in with views on the deliberation items. The leads come prepared with their owed deliverables. 45 minutes.

What failure looks like: Free-tier conversion economics get hand-waved. “We’ll figure it out post-launch” is the path to a $50K/mo Free-tier burn that subsidizes a 3% conversion rate.

Three decisions that need real deliberation

1. Free-tier circuit-breaker dial design

The ask: Approve product-tunable dials (chat cap, site-analysis cadence, digest model) that the Product Lead can turn without a deploy when Free-tier economics demand it.

Why this is yours, not engineering’s. The V1 cost discipline review surfaced that 50K Free × $1.07/mo = $53.5K/mo burn, while 5% conversion (the SPEC’s floor target) × 50K × $29 × ~70% margin ≈ $50K/mo gross. The Free-tier model breaks even AT the conversion floor, not above it. If we hit 8% conversion, the model works. At 3%, Free becomes a subsidy. That’s not an architecture-fixable problem — it’s a strategic-call about how protective the Free tier should be vs. how generous.

Trade-offs you’re weighing:

  • Tighter Free defaults protect margin but compress upgrade-moment surface area — and per your JD, conversion economics are load-bearing.
  • Looser Free + runtime dial preserves the conversion engine but means we ship with a cost gun on the table.
  • Hardcoded caps are simpler but every change becomes an engineering ticket, slowing reactive tuning.

My view if I were briefing you: Approve the dial pattern. Set conservative ship defaults (assume 3% conversion, not 5%). Require a written “what triggers tightening” rubric before Free signups open — not after we discover we’re underwater. (The quarterly free-tier review question is the same rubric, downstream.)

Cross-check refinements:

  • Product Lead: The rubric must include conversion-impact telemetry per dial setting, not just cost. Tightening the chat cap is an upgrade-moment surface change. Frame the rubric as joint cost+conversion, not pure cost defense. Owns the May 22 deliverable.
  • Lead Architect: The runtime dial means dial values live in the entitlement service config, not env vars — small schema addition folded into ADR-0001. Eval-regression also needs to be a rubric input (cohort cost can drop because eval pass-rate dropped — same number, different story).

Time pressure: Not Monday-blocking. Rubric on your desk by May 22.

2. Launch packaging — discount + existing-MP-customer trial + brand presentation

The ask: Three packaging calls that should land together — they reinforce each other:

Trade-offs:

  • Launch discount drives early conversion velocity but distorts the 60-day conversion read for the first cohort — and that metric is the most important number in our V1 economics dashboard.
  • 30-day trial for existing MP base is the cleanest “we appreciate you” gesture and exercises the upgrade-moment surfaces under real load before we lean on them at GA.
  • Brand: “MemberIntel” with Pro upsell is the cleanest story. “Insights Free” muddies our independent-product narrative and ties our brand fate to MP’s.

My view if I were briefing you: No launch discount. 30-day Pro trial yes for existing MP customers (one-time, opt-in, instrumented). Brand it “MemberIntel” — Pro is the upsell, not a separate product line. The discount adds noise to the metric I most need to read clearly in months 1–3.

Cross-check refinement (Product Lead): The brand call cascades into 3+ copy artifacts (pricing page, in-MP-admin banner, signup consent) that must comply with SPEC §4 honesty rules (no implication of training on non-customer MP data). Product Lead will produce a one-page “brand decision → copy obligations” map before launch comms drafting in Phase 4. You sign the map, not each artifact. The Free-tier opt-in consent copy is also gated by this — it’s not pure UX, it’s compliance + brand surface.

Time pressure: Not urgent Monday, but decide before Phase 4 launch comms drafting begins. Better now than under launch pressure.

3. ADR-0005 — Anthropic dependency mitigation seam

The ask: Sign ADR-0005. Approve the abstraction-seam approach as our Anthropic-dependency posture for V1.

Why this is yours. Per your JD §3 (Architecture material choices), you sign off on the Anthropic dependency and any plan to mitigate it. Seth signed the ADR on 2026-05-08 as an accepted risk; your signature is required for it to go operational.

Trade-offs:

  • Thin seam now (the existing llm.call wrapper is the seam) is one day of Seth’s time and preserves optionality. This is the May 5 decision in ADR form.
  • Going further (multi-provider routing at launch) buys insurance we don’t need yet and complicates eval.
  • Going lighter (no seam) puts us on a multi-week refactor if Anthropic raises prices, throttles, or has an outage we can’t ride out.

My view if I were briefing you: Approve. Same call as May 5. I want the seam, I want one written test that proves it works against a stub provider, and I do not want anyone building a second-provider integration before V1 GA.

Time pressure: Sign Monday. Per Seth’s cross-check: Slice 1 of the V1 product (first feature + GCP deploy) is otherwise unblocked by ADRs 0001–0004; ADR-0005 needs your Monday signature before Seth scaffolds the LLM client.

Confirmations seeking your sign-off

Brain content lead hire (carry-over from May 5 prep)

Recommendation from Seth + Product Lead: Hire dedicated brain content lead in May, separate from Sarah’s Phase 3 marketing/launch role.
My view: Approve. Same call as May 5. May 12 soft deadline still holds — Product Lead confirms Monday that sourcing has started.

Differentiation eval as Phase 3 milestone + monthly review

Recommendation: Adopt with 30 min/month executive review starting Phase 4. Per phased plan Phase 3 milestones.
My view: Approve. Monthly is the floor; Product Lead can pull me in more often if the gap narrows.
Refinement (Lead Architect): The release-gate threshold is Blair-shaped per Risk #1 mitigation — Seth brings methodology + proposed pass bar at Phase 3 milestone; you sign the bar. (See v1-eval-methodology.)

Infrastructure-leaning engineer on V1.5 hiring roadmap

Recommendation: Add to roadmap for V1.5 ramp (Q1 2027), not deferred to V2. Per V1.5 spec Phase 0 plan.
My view: Approve as a budget signal, not a today-commitment.

V1.5 trial-state schema fields in V1 entitlement service

Recommendation: Reserve trial-state fields in entitlement schema now (per ADR-0001) so V1.5 wizard doesn’t require migration.
My view: Approve — but Product Lead owes a one-pager on V1.5 trial flow shape (covers v15-trial-promo-existing-mp) before schema is final, so we reserve the right fields. Two-week deadline; outline brought Monday.

Quarterly free-tier review rubric (v1-quarterly-free-tier-review)

Recommendation: Approve once circuit-breaker dials are designed (downstream of Decision #1 above). Same rubric drives both.

LLM tracing tool — ADR-0006 (NEW)

Recommendation from Lead Architect: Self-hosted Langfuse on Cloud Run.
Why this is Blair-shaped: Vendor SaaS options (Langfuse cloud, Helicone, Phoenix) route user prompts + retrieved brain chunks through a third party — a data-egress decision that touches GDPR/CCPA posture, not just observability. Connects to v1-observability-stack.
My view: Approve self-hosted lean. Seth brings ADR-0006 skeleton Monday; sign by May 22.

Status updates from your direct reports

From Seth (Lead Architect)

  • ADR-0006 skeleton for LLM tracing — vendor shortlist + data-egress posture + recommendation. On your desk Monday.
  • Senior AI Engineer hiring status — sourcing pipeline count, current loop stage, target close date (mid-June close per JD success measures), two-line risk note. Visibility item, not a Blair decision; the most consequential Q2 hiring call is happening.
  • Cost-review numbers refreshed against ADRs 0001/0002/0003 — confirm $1.07 Free still holds with locked architecture; flag any drift. (Source: V1 cost discipline review.)
  • Slice 1 readiness check — first feature + GCP deploy. Unblocked by ADRs 0001–0004; needs ADR-0005 signature (above) to scaffold the LLM client.
  • Input-token ceiling proposed number — input to Product Lead’s circuit-breaker rubric. One sentence, one number.

From the Product Lead

  • Privacy counsel scheduling status — confirm half-day session is on calendar with counsel + Seth + Product Lead. If not, June 1 gate flag goes red Monday. Seth confirms he’s on the invite with prep doc (warehouse schema, retention, deletion pathway) ready 48 hours prior. (Companion: Privacy Counsel Architecture Review Agenda.)
  • Product Lead seat status — interim plan if role is still open, with the list of calls being provisionally made and which need re-confirmation post-hire. This is the most consequential dependency on Monday’s list — five of the seven May 5 confirmations and several Monday items assume a Product Lead is in seat.
  • V1.5 trial-flow one-pager — shape sketch so entitlement schema reserves correct fields. Outline Monday; full doc within two weeks.
  • Phase 1 scope-cut candidates — answer to “what comes out, not what slips” — Product Lead comes with 2–3 pre-identified cut candidates ranked by conversion-funnel impact. (Reference: Product Lead’s Phase 1 Deliverable Checklist.)
  • Circuit-breaker rubric outline — table of contents for the May 22 deliverable, so Blair sees the conversion lens is included alongside cost.

Decisions explicitly deferred (with triggers)

ItemSlug (linked)Trigger that reopens
Naming + trademark checkv1-naming-trademarkProduct Lead has marketing site copy draft. Not Phase 1.
Cross-pollination cadencev1-cross-pollination-cadenceBrain has 30+ playbooks. Until then, monthly is fine.
Customer support model for Free usersv1-customer-support-model1,000 Free signups OR first support escalation, whichever first.
Audit log retentionv15-audit-log-retentionCounsel architecture review (late May). Compliance-flag note (Product Lead): if counsel raises retention as a V1 obligation, this promotes to launch-gate.
BB launch date / partnership terms / admin UX / REST API / second content lead / multi-platform accounts / BB support modelv2-spec/* (7 items)V1.5 ships AND BB Memberships has confirmed launch date. Reopening any earlier is premature.

Resolved since May 5

ItemDecision artifactOne-line note
Hosting choice (v1-hosting-choice)ADR-0004GCP: Cloud Run + Cloud SQL + Secret Manager + BigQuery
Vector store (v1-vector-store)ADR-0003pgvector with tenant_id partition + RLS
Observability stack (partial) (v1-observability-stack)ADR-0004 + (ADR-0006 pending)BigQuery sink picked; LLM-tracing tool = Monday CONFIRM
Model routing source-of-truthADR-0002Single source of truth locked in entitlement service
Entitlement service shapeADR-0001V1.5 trial fields reserved (see CONFIRM list)
Anthropic dependency postureADR-0005Signed by Seth; awaiting your Monday signature

Routed to your leads (NOT MINE — engineering tactical or PL execution)

ItemOwnerNote
Token budgets per tierSeth + PLBring me a ceiling, not a number. PL flag: ceiling that implies Free chat-cap below 20/day becomes brand/conversion call, not engineering tactical — surface it.
Stripe Connect vs OAuthSethEngineering tactical, within material-choice envelope.
Eval methodologySethMethodology is Seth’s; release-gate threshold I sign at Phase 3 (above).
MP API surface for V1 syncSeth + MP teamEngineering coordination.
Free-tier opt-in flow UXPL → Blair signs disclosure languageNot pure UX — consent copy is brand+compliance. PL drafts; I sign disclosure language as part of brand decision.
MP MCP scopeSeth + MP teamConfirm at V1.5 build kickoff.
MP onboarding UX placementPL + MP productCross-team UX.
Card-required trial activation rate analysisPLData work, V1.5 Phase 0.
Eval suite ownershipSethEngineering org call.
Wizard error fallback (v15-wizard-mp-error-fallback) / re-run UX (v15-wizard-rerun)PLUX.
Agent rate limitsSeth + PLTune with usage data.
Content lead workload modelPLOperational.

Five things NOT in the inventory I’d add to the agenda

  1. Free-tier conversion-floor break-even (covered above as Decision #1) — the most important number on the table; not in any open-question doc previously. Source: V1 Cost Discipline Review.
  2. Privacy counsel late-May review confirmation (covered in PL status update) — if scheduling slips, June 1 gate slips and that’s a public call you have to make. Companion: Privacy Counsel Architecture Review Agenda.
  3. Product Lead role status (covered in PL status update) — if seat is unfilled, several Monday confirmations are provisional.
  4. “What are we cutting?” check on Phase 1 scope — five ADRs in one week is a lot of locked architecture. Standard scope-discipline question: anything in Phase 1 looking shaky on timeline, what comes out (not what slips). PL brings 2–3 candidates. Reference: Phase 1-2 Friction Points.
  5. LLM tracing tool pick (covered as CONFIRM #6 above) — only V1 observability piece still open after ADR-0004.

Reference shelf — companion docs

Document version: v1.1 — generated 2026-05-09 evening from the persona dev loop. Reviewed by Seth + Product Lead’s lenses before write. Updated 2026-05-09 with full link coverage to KB pages, ADR external links, and open-question anchors. Update with meeting outcomes after Monday.

For: B Blair Williams S Seth Shoultes P Product Lead S Santiago Perez Asis A AI Engineer