reference
Friction-Points Prep for Blair
Strategic prep document for Blair before the friction-points working session: three decisions that need real deliberation (brain content lead hire, differentiation eval cadence, infra-leaning hire on V1.5 roadmap), four confirmation items, predictable team disagreements, what to watch for in the meeting, and the asymmetric May-12 deadline on item 1.
Drafted: May 5, 2026
For: Blair Williams, CEO
Format: Read before the meeting. ~10 minutes.
What this meeting is, in your terms
A 30–45 minute working session with you, Seth, and the Product Lead (or whoever’s filling that role on the day) to land seven decisions that came out of the May architecture sessions. The decisions reshape (lightly) the phased plan and govern how Phase 1 executes starting May 4.
The meeting exists because the architecture work in May surfaced seven friction points where the original phased plan and the architectural commitments don’t quite fit. Some are budget calls. Some are calibration. Some are confirming recommendations Seth or the Product Lead is bringing to you for approval. None require you to learn new architecture.
What success looks like: seven decisions resolved (or explicitly deferred, with a date) by the end of the meeting. Phase 1 starts May 4 against an updated baseline that everyone agreed on rather than one that’s tentative.
What failure looks like: decisions get deferred to “a future conversation.” Phase 1 starts on a baseline the team is operating against tentatively. Mid-Phase 1, the same conversations reopen and slow Phase 2 prep.
Three decisions that need real deliberation
These are the ones where you should have a view before walking in. The other four (covered next) are essentially confirmations.
1. Brain content lead — hire for Phase 1, separate from Sarah?
The ask: Approve a new hire — a brain-authoring content lead — to start in May, not Sarah’s Phase 3 start.
Why it matters: The SPEC’s #1 risk is differentiation from generic AI. The differentiation moat is the brain. The brain quality at GA depends on how many high-quality playbooks the team has authored and how well retrieval surfaces them. The phased plan v1 had Sarah doing brain authoring + marketing/launch content + AEO + customer education + email campaigns + cross-pollination review. That’s three jobs braided together. Realistic outcome with one person: ~10 weeks of brain authoring before GA, landing the differentiation question on the SPEC’s optimistic case.
The recommendation: Hire a dedicated brain content lead in May. Sarah stays on her Phase 3 schedule for marketing/launch content. By GA there are two content roles. By V1.5 the SPEC v2 already calls for the second hire — this just pulls it forward by ~6 months.
The trade-off you’re weighing:
- One additional hire vs. accepting that V1’s differentiation lands on the optimistic case.
- The hire is the highest-leverage decision available for the SPEC’s #1 risk. No other change in the plan addresses it as directly.
- If the right person can’t be sourced quickly (which is real — content leads with MP domain knowledge plus privacy literacy plus editorial judgment is a thin candidate pool), the timing benefit erodes.
Time pressure: A “yes” by May 12 means the role can be sourced through May, with a candidate in seat by early June. A “yes” later than May 22 starts to lose the head start that’s the entire point.
My view if I were briefing you: Approve. The SPEC’s #1 risk is the one most likely to determine whether MemberIntel’s differentiation story holds at launch, and this hire is the single most direct lever on it. The cost is real but bounded. The downside of not hiring is not just slower brain growth — it’s a content lead role at GA that’s structurally overloaded, with all the cascading quality issues that creates.
3. Differentiation eval as a Phase 3 milestone + monthly review with you
The ask: Make the differentiation eval (MemberIntel scored vs vanilla Sonnet on MP-operator scenarios) a Phase 3 milestone, and commit to a monthly review with you starting Phase 4.
Why it matters: This is the leading indicator for the SPEC’s #1 risk. The eval shows whether the brain + retrieval are producing answers measurably better than what a customer would get from generic ChatGPT. If the gap isn’t widening by month 3 (end of Phase 3), the team has time to course-correct. If the gap is first measured at Phase 5, the only correction is “slip GA.”
The recommendation: Adopt as a Phase 3 milestone with monthly executive review.
The trade-off you’re weighing:
- Recurring time commitment from you — roughly 30 minutes once a month, starting Phase 4.
- The metric needs to be visible at the executive level, not just engineering, because the response to a narrowing gap is a strategic conversation (more playbooks? different retrieval? content lead acceleration? scope cut?), not just a technical one.
My view if I were briefing you: This is essentially a confirmation, but I flagged it as deliberation because it puts a recurring item on your calendar and that’s worth your conscious decision rather than passive acceptance. 30 minutes a month for the metric most predictive of the product’s strategic position is high leverage.
5. Infrastructure-leaning engineer on V1.5 hiring roadmap (not deferred to V2)
The ask: Add an infrastructure-leaning engineer to the hiring roadmap for V1.5 ramp (Q1 2027), rather than deferring to V2 (mid-2027).
Why it matters: Once V1.5 ships agents + greenfield wizard + the MP MCP integration, the system has materially more failure surface. Two engineers (Seth + Senior AI Engineer) absorbing ops on top of feature work is a burnout risk. The architecture has good defenses (kill switches, runbooks, observability), but those reduce ops load, they don’t eliminate it.
The recommendation: Plan for the hire post-Senior-AI-Engineer in V1.5 ramp. Specific timing and scope land later — this is a budget signal, not a Phase 1 decision.
The trade-off you’re weighing:
- One additional engineering hire ~9 months earlier than the SPEC’s “V2 reconsider” implies.
- Deferring to V2 saves the cost but increases the risk of Seth or the Senior AI Engineer burning out during V1.5 build, which has worse downstream cost.
My view if I were briefing you: Approve as a roadmap signal. You’re not committing to the hire today — you’re saying “this is on the roadmap before V2 and we’ll size it then.” The cost of saying yes today is zero; the cost of saying no is a harder budget conversation later when the load is already too high.
Four decisions that are essentially confirmations
These four have clear recommendations from Seth or the Product Lead with high-confidence rationale. Your role is approval, not deliberation. They take ~5 minutes total in the meeting.
2. Phase 2 milestone language calibration
The phrase “initial chat advisor working with Haiku model” in v1 of the phased plan oversells what’s realistic in 4 weeks with a brand-new senior hire ramping. Rev 2 calibrates: “initial chat skeleton — a route returns a Haiku response. Citations, eval, per-customer brain → Phase 3.”
Decision: Acknowledge the calibration. No scope change; just stop calling it “advisor” in Phase 2.
Why this is easy: It’s labeling, not planning. The actual work was always going to land where Rev 2 says.
4. Substantive privacy counsel architecture review in late May
V1 had counsel engaged by June 1 with compliance work intensifying in Phase 3. Rev 2 schedules a half-day working session with counsel + Seth + Product Lead in late May, before Phase 2 starts. Counsel sees the architecture before code is built against it, rather than reviewing post-hoc.
Decision: Approve the late-May counsel review and budget Seth’s prep time (one day).
Why this is easy: It’s an existing engagement scheduled earlier, not a new commitment. The cost is one day of Seth’s time. The benefit is preventing rework across Phases 2–4 if counsel surfaces architectural concerns. The agenda for the meeting is already drafted.
6. Model abstraction layer in Phase 1
A thin wrapper around the Anthropic SDK in Phase 1 — one day of Seth’s work — preserves the option to add a second model provider later as a config change rather than a multi-week refactor across every LLM call site.
Decision: Approve as a Phase 1 deliverable.
Why this is easy: Cost is one day. Benefit is real optionality (resilience, cost control, vendor leverage). Seth is the decider per the matrix; this is essentially you confirming his call.
7. Per-customer brain versioning in Phase 1 schema
Add a version column and superseded_at timestamp to per-customer brain entries from day one. Every update writes a versioned row; prior versions retained. The customer can see history and revert.
Decision: Approve the schema addition.
Why this is easy: Cost is 30 minutes of schema work in Phase 1. Benefit is preventing both an embarrassing customer experience (“the AI used to know X about me and now it doesn’t”) and a future migration if the team realizes after launch that versioning was needed. Seth is the decider per the matrix; you’re confirming.
Where the team might disagree (so you can prepare to tie-break)
The decision-rights matrix gives you final say on strategy and material architecture. In this meeting, the predictable tensions:
Seth and the Product Lead may disagree on item 1 (brain content lead hire). Seth’s perspective is technical: “the brain is the moat, we need depth, hire whoever it takes to make it deep.” The Product Lead’s perspective tends to be cross-functional: “I have other content needs, hiring two people in May spreads the team thin.” Both are right from their angles. Your tie-break is strategic: which perspective serves the SPEC’s #1 risk better. (Recommendation: Seth’s framing wins here, but for product reasons not technical ones — the brain is the differentiation lever.)
Seth may want to push item 5 (infrastructure hire) earlier than V1.5 ramp. If V1 launch is rough, Seth might come back in Phase 6 saying “I needed this hire two months ago.” Your tie-break: hold the V1.5-ramp commitment for now, but acknowledge that if Phase 5 launch surfaces serious ops issues, the hire moves up. The signal you’re sending isn’t “definitely V1.5”; it’s “before V2 and probably V1.5.”
The Product Lead may want item 3 (differentiation eval review) more often than monthly. If the differentiation gap looks shaky in Phase 3 or Phase 4, more frequent review is reasonable. Your tie-break: monthly is the floor; the Product Lead can request additional time if the metric trends concerning. Don’t lock to a fixed cadence; lock to a minimum.
Item 7 (brain versioning) is technical enough that the Product Lead might not have a view. That’s fine. It’s Seth’s call confirmed by you; the Product Lead’s role is to surface any product-experience implications (customer-facing UI for brain history, support implications), which they can flag if relevant.
What to watch for in the meeting
Drift toward “let’s discuss this later.” If two or more of the seven decisions get deferred without an explicit follow-up date, the meeting hasn’t accomplished its purpose. Push for resolution-or-explicit-deferral on each.
Item 1 turning into a hire-spec conversation. The decision is yes/no on the role existing. The decision on who and the role spec is downstream. If the meeting drifts into “what should the JD say,” redirect — that’s a follow-up the Product Lead and Seth can handle.
Items 2, 4, 6, 7 getting more time than they need. They’re confirmations. If 30 minutes goes by on items 2 and 4, the deliberation items haven’t gotten their fair share.
Optimism on Phase 2 scope. Even with item 2’s calibration, there will be a temptation to push the milestone language back toward “we can probably get more done.” Your role is to hold the calibrated version. Seth or the Product Lead may inadvertently overcommit; you under-promise so they can over-deliver.
Decisions that imply budget without saying so. Items 1 and 5 are budget calls. The meeting needs to acknowledge this explicitly so the budget impact is on your desk, not surfaced quietly to finance later.
What slips if decisions don’t land
For your own clock-watching:
- Item 1 not landing by May 12: brain content lead can’t be sourced in time to be productive before GA. Differentiation depends on Sarah’s bandwidth alone. SPEC’s #1 risk lands on the optimistic side.
- Item 4 not landing by May 12: counsel architecture review can’t be scheduled by late May. Counsel sees architecture for the first time in Phase 3. Rework risk in Phase 2.
- Items 2, 6, 7 not landing in this meeting: Phase 1 starts on a baseline the team is partially operating against; reopens later in Phase 1 or early Phase 2.
- Item 3 not landing: differentiation gap first measured at Phase 5 (launch ramp). Course-correction window lost. Possible GA slip if measurement reveals weakness.
- Item 5 not landing: budget signal absent. Hire conversation reopens in Phase 6 under more pressure, or in V2 too late.
The asymmetric one is item 1. Every other decision can slip a week or two without serious consequence. Item 1’s calendar is unforgiving — May 12 is the soft deadline, May 22 is the hard one.
Quick decisions cheat sheet
If you scan during the meeting:
| # | Decision | Recommendation | If approved | If denied |
|---|---|---|---|---|
| 1 | Brain content lead hire in Phase 1 | Approve | One additional hire; SPEC #1 risk addressed | Risk lands on optimistic case; rework likely |
| 2 | Phase 2 milestone calibrated | Approve | Honest expectations | Phase 2 ships against optimistic baseline |
| 3 | Differentiation eval as Phase 3 milestone + monthly review | Approve | 30 min/month from you starting Phase 4 | Course-correction window lost |
| 4 | Counsel architecture review late May | Approve | One day Seth prep; preempts rework | Counsel sees arch in Phase 3; possible rework |
| 5 | Infra hire on V1.5 roadmap | Approve | Budget signal only | Hire conversation under pressure later |
| 6 | Model abstraction layer in Phase 1 | Approve | One day Seth’s work; preserves optionality | Multi-week refactor if option ever needed |
| 7 | Brain versioning in Phase 1 schema | Approve | 30 min schema work; prevents future migration | Migration work + customer comms moment later |
Note on the Product Lead role transition
The friction-points one-pager and Phase 1 milestone gate criteria assume the Product Lead role is filled when the meeting happens. If the role is in transition and the meeting happens before the new Product Lead is in seat, two adjustments are worth thinking about:
- You and Seth can resolve five of the seven decisions (items 2, 4, 5, 6, 7) without a Product Lead present. They’re either confirmations or your-call items.
- Items 1 and 3 benefit from Product Lead input because they shape product-side workstreams. If the role is open, you can either defer those two specifically until the new hire is in seat, or make provisional decisions with Seth and confirm with the new Product Lead within their first two weeks.
The cleanest path is to land the meeting with whoever’s filling the role on the day — even an interim — and treat items 1 and 3 as provisional if necessary. Drifting the meeting until the role is permanently filled risks the May 12 deadline on item 1.
Drafted by: Seth Shoultes, with Product Lead input
Read by Blair before: [target meeting date]