What we expect to be asked, and how we answer it.

Is "biomarker-informed financial planning" a real category or a marketing frame?

Real category with three converging inflections that did not exist as recently as 2024. Data: Function Health at 100K+ paying members and a $2.5B Series B put personalized biomarker inputs at consumer scale for the first time. Distribution: concierge medicine now serves ~300–500K HNW patients in chains with existing longitudinal biomarker relationships. Delivery: frontier LLMs are now capable of the bounded tool-use + structured-output patterns required to ship a regulated consumer decision-support agent. No incumbent occupies the layer that turns biomarker data into financial-planning parameters — HIFP is defining it.

Why hasn't anyone else built this? Isn't the market gap suspicious?

The gap exists because the three inputs converged only recently, and the closest incumbent (Genivity/HALO, acquired by Lumiant April 2023) has contracted to 7 employees as of July 2026 with no capital since March 2023. The category was too early three years ago and now looks obvious in retrospect. First-mover pain is real (novel regulatory jurisprudence — see priced-risk section) but so is the 12–18 month competitive window before a serious player could ship a comparable product.

What's your one-line positioning?

"HIFP is the biomarker-native financial planning platform for the decumulation decades — physician-distributed, actuarially grounded, and delivered through a personalized planning agent gated by a published Advice-Boundary Architecture no competitor can replicate at speed." The category is biomarker-native planning; AI is a delivery mechanism, not the identity.

Is the HNW-only focus a TAM ceiling that caps the exit?

No. Serviceable segment is ~5M US households with $1M–$25M investable assets, plus ~300–500K concierge patients, plus F500 executive populations. Blended path to $60M+ ARR by Y4 across four consumer/enterprise revenue streams. The HNW anchor is a wedge, not a ceiling — the freemium tier captures broader top-of-funnel volume for optionality and eventual mass-market extension. TAM math and revenue trajectory are laid out in the pro-forma; not a cap.

How do you get to freemium free-to-paid CVR ≥6% at 90 days? That's aggressive.

The freemium funnel is engineered for conversion, not casual signup. Cold subscription CVR is not the tractable optimization; free-to-paid CVR is. Gate 4 documents the metric explicitly as falsifiable — if we hit 3–6% we extend the timeline; if we hit <3% we reconsider the D2C-only path. The gate is designed to fail loudly. Multi-modal onboarding ingestion (20-min → 3-min setup) is the specific CVR-lift mechanism, plus co-marketing with Function/Superpower and a concierge-chain warm-share pattern.

Isn't Function Health going to build "Function Plan" and eat you?

Function is the co-marketing partner, not the competitor. Their next product bet is more likely to be in adjacent verticals (women's health, pediatric longevity, clinical-trial recruitment) than in decision-support planning — those verticals are closer to their competence and higher-margin than a rules-based planning tool. If Function did decide to ship a competing planning agent, the actuarial model + Advice-Boundary Architecture + regulated-industry compliance discipline is a build they have not started. That's our lock-in window.

"AI Planning Agent" — how is this different from Wealthfront-AI or SigFig?

Fundamentally different in three ways. (1) Grounded in biomarker data + actuarial model, not general planning heuristics — Wealthfront optimizes portfolios; HIFP produces care-reserve trajectories from your Function panel and Oura data. (2) Bounded architecture — the planning agent is a tool-use layer over ~15 named parameterized functions, not a general-purpose LLM chat. (3) Gated by a published Advice-Boundary Architecture — a compliance layer no AI-fintech incumbent has invested in because they haven't needed to. HIFP does need to, because we sit at the intersection of health + finance + AI where all three regulatory bodies apply.

Why isn't AI a fifth pillar? Why "connective tissue"?

Because AI as a fifth pillar concentrates the AI story into a single organizational silo — which is the opposite of how you ship AI in a regulated product. AI is properly the delivery mechanism for the actuarial model (Pillar 1), the translation layer (Pillar 2), and the trust posture (Pillar 4). "Connective tissue" reflects the architecture; "fifth pillar" would misrepresent it.

Why not build on Salesforce Agentforce? You're a Salesforce person.

Two reasons Agentforce is the wrong hosting layer for the consumer Planning Agent. (1) Per-user licensing kills consumer freemium — Agentforce is priced per user + per action; the Longevity Snapshot free tier and Plan Core at $9–19/mo cannot sustain those economics. (2) We lose control of the Advice-Boundary Architecture — the entire regulatory-defensibility story depends on us owning the tool whitelist, template library, classifier, and audit log end-to-end. Agentforce abstracts away exactly the layers we need to control. Salesforce earns its place elsewhere in the stack — Sales Cloud for internal CRM at Y2 — but not here.

Why Python + TypeScript instead of one language?

Python for the actuarial + ML + agent orchestration layer (that's where the numerical stack lives — numpy, pandas, scipy, statsmodels, PyTorch/TF for classifier). TypeScript at the edge for auth + rate-limit + audit-tap latency-critical work, and for the consumer front-end (Next.js + PWA at MVP; React Native mobile in H1). Two languages, one for each domain of expertise — reduces team size vs. forcing a Node-native ML stack or a Python-native edge stack. Standard modern architecture for a health-AI startup.

WA MHMDA class-action exposure is huge. How are you protected?

Ten-move architectural mitigation, plus explicit strategic timing choice: Planning Agent is geo-gated to non-WA + non-IL residents at MVP. H1 expansion to WA + IL only after 6 months of ex-jurisdiction production evidence. This retires ~85% of the novel-jurisprudence risk at the cost of ~4% of TAM. Consent-scope as a first-class object, jurisdiction-aware consent flow, cyber-insurance floor, and specialist counsel (Manatt or Perkins Coie) engaged month 1. Documented as priced risk, not hand-waved.

The planning agent will hallucinate. What happens then?

Two-layer safety net + fail-closed defaults. Generation-side: structured intent + template rendering means the LLM never emits free-text to users — it selects a template and fills it with parameters from the deterministic model. Runtime-side: every rendered output passes an Advice-Boundary Classifier before display; failure → scripted deflection + 24-hour incident review + regression test added. Release-gate: adversarial red-team test set of 500+ prompts, growing to 5K+ in H2; zero-tolerance failure classes (free-text financial recommendation, diagnostic claim, consent violation, cross-user leak) at 0% failure rate; bounded-tolerance classes at ≥98%. If the gate doesn't clear, we ship a v2-shape MVP (no Planning Agent Q&A) — documented fallback.

Do the Anthropic / OpenAI / Google acceptable-use policies actually allow this use case?

Yes, with architectural discipline documented as evidence. Anthropic (our MVP primary) explicitly classifies healthcare + financial decisions as high-risk requiring "qualified professional review... prior to dissemination or finalization" and "disclose to end users that you are using AI." HIFP's compliance interpretation, validated by the Advice-Boundary Architecture design: outputs are planning parameters not advice/decisions/recommendations, so the qualified-professional-review requirement is met at the tool-design layer (ASA-LTC lead, Sindhu as clinical anchor, Fatima as CEO, and specialist counsel review the tool whitelist + templates + classifier taxonomy quarterly) rather than per-output. User disclosure is met by first-session attestation + persistent UI affordance + ToS language. Anthropic enterprise agreement pursued in H1 to memorialize this interpretation contractually. Google AUP is materially weaker; HIFP compliant. OpenAI manual verification pending; not MVP-blocking (Anthropic is primary).

What about SEC / state RIA advice-boundary enforcement?

The architectural red line is bright: outputs are planning parameters, never recommendations. "Reserve $180K in years 6–10 with 78% confidence" is a parameter. "Buy this LTC policy" is a recommendation. Every user-facing string is scanned by the Advice-Boundary Classifier against SEC/state RIA case law before display. Counsel signs off on the tool whitelist + template library per release. Novel LLM-in-loop SEC enforcement risk exists (SEC has not brought a public action yet but is signaling intent through the 2025–26 Marketing Rule cycle) — priced explicitly as residual risk; response playbook + PR readiness maintained.

FDA SaMD — how are you not a diagnostic device?

Same architectural red line applied to clinical claims. "78% probability of cardiac event" would be a diagnostic device claim (regulatory kill-shot). HIFP surfaces planning parameters that consume biomarker inputs — not clinical predictions. Compliance review on every user-facing string as release gate. Physician-facing cohort dashboard shows planning aggregates (reserve trajectories, downsize-timing distributions), never clinical aggregates (LTC-onset probabilities by cohort) — a product-review rule that keeps the dashboard on the right side of the SaMD line.

Two of your three founders are "prospective." That's a huge caveat. Why should we invest in a team that isn't fully assembled?

Because pretending it's closed damages credibility more than acknowledging it costs. HIFP is Sindhu's founding concept, her equity anchor, her clinical thesis — she is the founder in every sense. Fatima and Brandon are prospective founding executives who close specific DD questions the founding team otherwise cannot answer: Fatima's health-strategic distribution + CEO gravitas, Brandon's regulated-AI product execution + Health Cloud + FinTech precedent. Both are pending — Fatima on Gate 1 full-time commitment (targeted Q2 2027); Brandon on founder-agreement close (his own 90-day framework in parallel). The v3 raise structure is designed so Series A closes AFTER Gate 1 and founder agreement — investors are not being asked to fund a fully assembled team on faith; they are being asked to fund a fully assembled team when it is fully assembled. If either commitment fails, HIFP either declines or restructures. That is how founder gates should work.

Fatima has a day job. When does she come full-time?

Gate 1: written founder agreement by 31 Jan 2027 with named transition date no later than Q2 2027. Ambiguous outcomes are treated as failure — the memo holds the criteria loudly. If Gate 1 fails, everything downstream defers or declines. This is the single biggest go/no-go signal in the whole plan.

Brandon holding combined CPO/CTO is a lot for one person. Single point of failure risk.

Priced and mitigated. Combined CPO/CTO through Series A is deliberately audacious — it also removes a $500–800K/yr hire from Y1 use-of-funds. Mitigation: staff+ Founding Engineer #1 hired at MVP-team formation, promotable to VP Eng in Y2. Net Y1 cost delta: +$180K vs. current v3 plan. Brandon's dual Salesforce Certified Architect (System + Application) + Salesforce AI Associate + current Agentforce Partner Innovation Lab lead is the credential floor that makes combined title defensible; Founding Engineer #1 is the operational safety net. Y2 VP Eng hire trigger documented.

Can a team of 4–6 engineers actually ship this in 12 months?

Yes at the scope-cut version documented explicitly. Ships at MVP: Q&A planning agent, Plan-Delta monitor, multi-modal ingestion, 4 consumer tiers, chain 1 integration, actuarial model + 2 back-tests, Advice-Boundary Classifier, SOC 2 in-flight. Deferred to H1: Brief mode, Roadmap mode, native iOS/Android, curated directory, share-with-incumbent, multi-provider LLM, physician cohort brief. Sindhu is aligned on the smaller-MVP posture — positioning holds, scope cuts by ~30%, timeline holds.

Where's the ASA-LTC actuarial lead? That's a Tier 1 gap.

Named as Tier 1 in the gap list; hire in flight in Phase 0 (pre-close). Signed offer + start date + publishable bio is the closed-when criterion. Fatima's network provides the candidate pool; Brandon assesses technical fit. Failure mode: hiring a data scientist without actuarial credential and expecting insurer conversations to advance in Y3 — explicitly guarded against.

Will a concierge chain actually pay $10–15/patient/mo?

Gate 6 is designed to falsify this exact assumption. 3+ chain-executive conversations at VP+ level with pricing discussed, ≥2 of 3 indicate the range is defensible. If not, we revisit either the chain-vs-boutique strategy or the pricing model itself. Chain economics differ materially from boutique — this is why the anchor MSA (Gate 2) is a chain deal, not a boutique deal. Fatima's current Oracle Health + former Salesforce H&LS + McKinsey + Microsoft networks reach MDVIP, One Medical Premium (via Amazon), Executive Health Group, PartnerMD at the buyer level directly.

Concierge sales cycles are 6–12 months. What if Gate 2 fails?

Two documented backstops. Shadow pipeline: 5–10 boutique practices signed as pilot MSAs run in parallel as early revenue + case-study fodder. Variant C: B2B-only insurer-first pivot documented as fallback; different bidder pool, different regulatory profile, narrower TAM but higher per-deal value. Not a psychological fallback (option-of-last-resort) — a real Variant with named acquirers.

Blended D2C paid ARPU at $22/mo — Boldin is $10, Function is $42. Are you sure that's right?

Yes. Blended $22 reflects Plan Core ($14 midpoint) + Plan Premium ($40 midpoint) weighted 60/40 in favor of Core at scale. Freemium funnel disproportionately captures price-sensitive users converting into Core; Premium mix improves with concierge co-marketing overlap. Upside case (heavier Premium mix) blends to $30 and lifts D2C revenue ~35%. Downside case (Core-dominant) is what the base pro-forma models.

$7–9M — why this size? Why not $10M like v1, or $6–8M like v2?

v1 assumed a headless-API-into-RIA thesis with a broader use-of-funds. v2 tightened to concierge-anchored + D2C with a smaller founding team (6–8) at $6–8M. v3 lands at $7–9M — the tightened v2 base $7M plus two disciplined additions: $400–600K contingency buffer (chain-integration surprises, adversarial-test-set curation partner slippage, novel-jurisprudence legal reserve — priced explicitly rather than absorbed by underfunded reserves) and $500K–1M growth acceleration option (deployable in H1 only if freemium CVR beats Gate 4 target). Use-of-funds sizes to actual burn: ~$2.3M Y1 OpEx (unchanged from tightened v3), ~26–28 month runway at $8M midpoint. Series B triggered at chain-live + Variant A opening in H2.

Health/longevity-strategic lead — why not a fintech generalist?

Different room, different diligence, more appropriate exit thesis alignment. Health/longevity strategics (a16z Bio + Health, ARCH Venture, GV, Lux Capital, health-strategic vehicles) understand the regulatory posture, the concierge distribution, and the biomarker-informed thesis; fintech generalists diligence on advisor-tech models we don't want to be. AI-forward generalists (Sequoia AI-application, Radical, Bond, Lightspeed AI) opened as a legitimate second lane — not swap, add. Wider room, better price-discovery.

The Y3 exit valuation ranges look aggressive. Are you sure about the AI-native strategic pool?

Base ranges (health-strategic, wealth-platform, insurer carve-out) are quantified from comparable transactions. AI-native strategic pool (Microsoft H&LS, Google Health, AWS HealthLake) is a narrative asset that requires DD-stage evidence to defend. Retained in the strategic narrative; flagged as directional in the exit valuation table so we're not caught pretending it's guaranteed. If we get to Y3 with the published Advice-Boundary Architecture + production shipping data, the pool becomes real; if not, we don't oversell it.

What's your defensible confidence in the v3 thesis?

0.64. A modest upgrade from v2.1 (0.62), justified per-delta: founder-team correction +0.03, product-design lock +0.01, AUP research clears T3 extinction risk +0.02, Planning Agent + WA/IL geo-gate residual -0.02, combined-role SPOF with FE#1 mitigation -0.02. Net +0.02. The v3 draft claimed 0.68; the Red Team v3 adversarial pass caught the overclaim and corrected. Honest confidence is a raise asset, not a liability.