HIFP · Direction / Pivot Change Log · Canonical
What HIFP decided, when, and why — so nobody has to re-ask.
Purpose + how to read
Standing record of material directional decisions across HIFP versions.
Prevents the "same question re-asked at every version" pattern surfaced in the Sindhu 2026-09-03 session — a new stakeholder can read this document and skip re-litigating settled questions.
Scope discipline. In: architecture, business-model, positioning, commercial-lever, partnership-class decisions. Out: refinements (copy, wording, tier pricing tweaks, artifact reorganization). Those live in artifact revision histories, not here.
Each entry states what changed, when, why (the specific driver), what it replaced, and downstream artifacts touched. A pivot with no downstream is not a pivot — it is a hallway conversation.
§1 · Architecture pivots
1.1
Headless CMS / headless architecture — considered then rejected
Whenv1 → v2 (2026-Q2 window)
What changedEarly v1 sketches contemplated a headless-CMS front-end + decoupled service layer for the marketing site + product surface.
Why rejectedTwo-team overhead at a stage where the eng team was 0–1 people; premature abstraction; the actual eng lift was in the actuarial + advice-boundary + data-ingest layers, not the CMS. Headless would have front-loaded infrastructure work that added no user value in MVP.
Replaced withSingle Next.js app for POC and MVP; Sanity CMS as headless marketing-site backer only once a marketing team exists to justify it (POV v3.1 §12).
DownstreamPOV v3, Product Design v3, Marketecture v3.
Revisit triggerSeries B — if front-end team ≥3 and marketing content volume warrants.
1.2
Cloud primary — GCP evaluated → AWS locked
Whenv3 → v3.1 r2 (2026-09-02)
What changedGCP considered (BigQuery + Vertex ML strength + healthcare-friendly BAAs). Locked AWS.
Why AWS won(a) HIPAA BAA maturity + chain-partner inspectability, (b) founding-engineer market depth on AWS is materially deeper, (c) SOC 2 tooling ecosystem (Vanta/Drata) is AWS-native-first.
Replaced withAWS primary; Cloudflare edge (CDN + WAF + Turnstile + DDoS); frontend on AWS Amplify or Vercel per founding-engineer preference.
DownstreamPOV v3.1 r2, Marketecture v3.1, Security Architecture v0.1, Product Design v3.1.
Revisit triggerMulti-region regulatory need (EU) that changes the calculus.
§2 · Business-model pivots
2.1
White-label delivery — considered then narrowed
Whenv2 → v3
What changedEarly consideration of white-label HIFP as the primary chain-partner delivery mechanism (chain-branded consumer surface, HIFP under the hood).
Why narrowedWhite-label breaks the trust-brand thesis (Pillar 4). If the value prop is "biomarker-informed planning with an advice-boundary you can trust," the brand IS the trust. White-label dilutes that at the exact moment the user needs it most.
Replaced withCo-branded surface with visible HIFP brand + chain partner; share-with-incumbent flow for advisor-side visibility. Variant A (OEM, Y2) preserves the white-label option as a downstream commercial path once brand equity exists.
DownstreamPOV v3 §10 (Variants), Chain Partnership Playbook v0.1.
Revisit triggerChain partner insists on white-label as a signing condition — evaluate case-by-case, do not default in.
2.2
Embedded model — considered then reshaped as share-with-incumbent
Whenv2 → v3
What changedEmbedded model contemplated — HIFP as widget/module inside advisor tech stacks (eMoney, RightCapital, Orion) or chain member portals.
Why reshapedFull embed cedes UX control and dilutes the actionable-value story. Advisor-tech vendors also carry their own advice-liability posture that would leak onto HIFP.
Replaced withShare-with-incumbent full flow (POV v3.1 §4) — HIFP-hosted surface with a public /share/preview/[id] advisor-side view + eMoney/RightCapital JSON exports. Preserves the "we don't compete with the advisor" signal without cannibalizing HIFP UX.
DownstreamPOV v3.1, Product Design v3.1, Financial-Baseline Addendum v0.1 (advisor-push tier).
Revisit triggerAdvisor-tech partner offers full-fidelity embed with UX + liability parity.
2.3
Commercial-lever count — four → five (viral) → six (data-source marketplace, candidate)
Whenv3 → v3.1 (four → five); v3.1 → v3.2 (five → six candidate)
What changedv3 → v3.1: added viral referral loop as 5th lever. v3.1 → v3.2 (proposed): data-source partnership marketplace under evaluation as 6th commercial lever (Session 2026-09-03).
Why the additionsPOC surfaced $42 reciprocal referral credit as naturally shareable in HNW 45–70. Data-source partnerships (Oura, Galleri, HLI, aggregators) surfaced in the Sindhu session as potentially high-signal CAC-reduction or revenue lever.
DownstreamPOV v3.1 §6, Financial Model v0.1 (revenue-mix sensitivity), Gate 7 addition (viral k-factor ≥ 0.15).
Revisit triggerGate 7 fails (viral drops to H1 experiment); or data-source partnership BD returns hostile.
§3 · Data-partnership pivots
3.1
Financial-data ingestion — Plaid/MX/Yodlee direct → tiered hybrid
Whenv3 → v3.1 r2 (2026-09-02, per Financial-Baseline Addendum v0.1)
What changedPlaid/MX/Yodlee direct-integration considered as the primary financial-data path. Rejected in MVP.
Why rejected(a) HNW 45–70 is the most Plaid-resistant demographic in retail finance — session-ending friction; (b) GLBA + state financial-privacy compliance surface would collide with WA MHMDA + IL GIPA in the same MVP quarters; (c) architectural red-line parallel — "we don't hold your money and don't need direct account access" is a pitch asset.
Replaced withTiered hybrid — Self-report buckets (MVP) + Advisor-push (MVP concierge) + Kubera aggregator (H1) + Advisor-tech reads (H2) + Empower opportunistic (H2). Plaid deferred indefinitely.
DownstreamFinancial-Baseline Addendum v0.1, Product Design v3.1 Data page, POC /data restructure.
Revisit triggerKubera collapses AND advisor channels stall AND MHMDA/GIPA posture matures — none expected in 24-mo window.
3.2
Large-fin platform partnerships (Envestnet-class) — pivoted away
Whenv2 → v3
What changedEnvestnet-class TAMP + advisor-tech infrastructure platforms considered as distribution partners.
Why rejected(a) Enterprise BD cycle (12–24 months) incompatible with pre-Series A timeline, (b) TAMP incentive structure funnels to their own model portfolios — competitive tension with HIFP's neutral posture, (c) contract shape typically includes non-competes that would block Y3 Variant B institutional data product.
Replaced witheMoney + RightCapital + Orion + Redtail as H2 BD targets (advisor-tech, not TAMP) — smaller integration surface, cleaner incentive alignment.
DownstreamFinancial-Baseline Addendum v0.1 §3.4, POV v3 §10 Variants.
Revisit triggerPost-Series B, if HIFP has traction leverage that inverts the negotiation.
3.3
Fintech-native partnerships (SoFi / Wealthfront / Betterment / Robinhood Gold) — evaluated, re-pivoted away
Whenv3.1 → v3.2 (Session 2026-09-03)
What changedFintech-native platforms evaluated as an alternative to the large-fin pivot. Question: does the Envestnet-class reasoning apply here, or does the fintech shape change the calculus?
Why re-pivoted away(a) Demographic dissonance — fintech-native user bases skew 25–45; HIFP target is HNW 45–70. Distribution partnership signals wrong demographic to chain partners. (b) Chain-channel signal mismatch — MDVIP / One Medical Premium members do not overlap with SoFi / Wealthfront members in meaningful volume. (c) Product philosophy tension — most fintech-natives compete on price-per-basis-point; HIFP competes on trust + actionable insight. Co-marketing signals commoditize HIFP.
Replaced withNo partnership pursued. Fintech-native path remains available as a Y3+ evaluation once brand equity exists.
DownstreamPOV v3.2 (planned), Partner Pipeline memo update.
Revisit triggerA fintech-native launches an HNW 55+ product (Wealthfront has hinted at $500K+ tier) that overlaps HIFP's demographic — evaluate specifically, not the class.
3.4
Fin-services product marketplace (product-referral shape) — considered then rejected
Whenv2 → v3 (implicit); made explicit in v3.2 pending founder-team ratification
What changedCurated product marketplace evaluated — insurance products, annuities, direct-indexing sleeves, actively-managed sleeves. Would have added a referral-fee revenue lever.
Why rejected(a) A curated marketplace is an implicit recommendation → collides with the Advice-Boundary posture, (b) referral-fee revenue muddles fiduciary posture even though HIFP is not a fiduciary — appearance of conflict is enough to break the trust-brand, (c) chain-partner conflict — advisors expect to be the marketplace; HIFP competing with them breaks Variant A, (d) share-with-incumbent tension — the promise to advisors is "we won't compete with you."
Replaced withNothing revenue-side. On the input side (data-source partnerships) the marketplace concept lives — see §2.3.
DownstreamPOV v3.2 (planned), Q&A v3.2 (explicit "why not a marketplace" answer).
Revisit triggerPost-Series B, only if the Advice-Boundary posture is materially reformulated to accommodate curated recommendations under a compliant framework.
§4 · Founder-team pivots
4.1
Founder-team composition — locked at three
Whenv2 → v3
What changedTwo-founder (Sindhu + Fatima) briefly contemplated in early v2 with Brandon as advisor. Three-founder (Sindhu + Fatima + Brandon) locked in v3.
WhyBrandon's regulated-AI + platform-integration + Health-Cloud precedent closes three DD questions in one seat (see hifp-founder-profile). Removes CTO hire from Y1 (~$500–800K savings). Investor room reads materially different with a three-corner founding team.
Replaced withThree-founder team. Brandon = Co-founder & Chief Product / Technology Officer (combined CPO/CTO through Series A → CPO after VP Eng hire in Y2).
DownstreamPOV v3, Founder Bio Correction v3, Org & Hiring Plan v0.1, Funding Path v3.
Revisit triggerFounder-contingency scenarios (Task 3.1 open) may model 2-of-3 or 1-of-3 outcomes.
4.2
Salesforce role — evaluated as partner / customer / exit-path; landed neutral
Whenv1 → v3, iteratively
What changedSalesforce role has been considered as (a) potential health-cloud data-partner, (b) potential customer (HIFP inside Salesforce H&LS), (c) potential acquirer.
Why landed neutral(a) as partner — Salesforce Ventures kept as a Series A source (Fatima + Sindhu + Brandon all Salesforce-adjacent), but no data-partnership dependency on Salesforce, (b) as customer — no product path routes through Salesforce H&LS, (c) as acquirer — kept out of the artifact framing entirely; acquisition is a downstream outcome, not a strategy.
Replaced withSalesforce Ventures = one of ~5 first-check sources. Fatima's + Sindhu's + Brandon's Salesforce relationships are network assets, not strategic dependencies.
DownstreamFunding Path v3, POV v3 §3.
Revisit triggerSalesforce approaches with a specific partnership (unlikely given org movement — Fatima now at Oracle, Sindhu now at Hyro).
4.3
Current-employer descriptors — Fatima updated (Oracle); Sindhu updated (Hyro)
WhenFatima 2026-09-02 (POV v3.1); Sindhu 2026-09-03 (v3.2, this pass)
What changedBoth co-founders left the Salesforce H&LS positioning that anchored earlier drafts. Fatima now at Oracle in a senior health-leadership role. Sindhu now VP, Clinical Product Strategy at Hyro (conversational AI in healthcare).
Why it mattersSoftens the "founding team IS the Salesforce H&LS incumbent" positioning of v3.1 r1. On Sindhu's side, Hyro strengthens the "AI + clinical + healthcare" credential in a way the Salesforce role did not.
Replaced withBio references + pitch framing updated across POV, Investor Deck, Infographic, Q&A, Funding Path, Verification Runbook, Source Log, and this Log.
DownstreamEvery raise-facing artifact.
Revisit triggerFurther role changes; both co-founder LinkedIns confirmed stale.
§5 · Product-scope pivots
5.1
Advanced Diagnostics — cut from MVP → added back as onboarding tile (H1-gated ingestion)
Whenv3 → v3.1
What changedAdvanced Diagnostics (Galleri, HLI, CAC, DEXA, polygenic) originally cut from MVP for scope discipline. POC surfaced the "why can't I add my Galleri result?" objection immediately.
Why the compromiseFull ingestion into the risk model in MVP would front-load IL GIPA + WA MHMDA scope review. Onboarding tile with H1-gated ingestion preserves scope while removing the objection.
Replaced withOnboarding-tile-only in MVP with "H1 · beta" tag and heightened-consent copy; ingestion activates in H1.
DownstreamPOV v3.1 §4, Product Design v3.1, Compliance Foundation v0.1.
5.2
Household mode — deferred to H1 → MVP tile stub + H1 full
Whenv3 → v3.1 r2
What changedHousehold mode originally H1-only. POC surfaced sandwich-generation as the highest-signal product insight.
Why the compromiseFull multi-user permission model + household pricing + compliance sub-spec is real H1-scale eng. Tile stub in MVP unlocks the story without shipping the complexity.
Replaced withMVP = view-only invite affordance under Settings → Household. H1 = full permission model + sandwich-generation surface + household pricing.
DownstreamPOV v3.1 §4, Product Design v3.1, Compliance Foundation v0.1 (household-consent sub-spec).
5.3
Gate list — six → seven (viral k-factor added, gating)
Whenv3 → v3.1 r2
What changedSixth gate added (viral k-factor ≥ 0.15 by day 90), initially supplementary, promoted to gating in r2.
Why gatingA passing k-factor unlocks the Series A viral-CAC narrative and validates the shift from four to five commercial levers. Soft signal is not enough — the raise story hinges on it.
Replaced withSeven-gate day-90 decision matrix (7/7 = join full-time, take Series A full pace; 6/7 = join, 30-day extension; 5/7 = extend full decision by 30 days; ≤4/7 = decline founder role).
DownstreamPOV v3.1 §7, Value-Decision Memo v3 (planned).
§6 · Ledger discipline
Adding a row. Any material directional decision — architectural, business-model, positioning, commercial-lever, partnership-class, founder-team, product-scope — adds a row. If in doubt whether a decision is directional or refinement, err toward adding it.
Retiring a row. Rows are never retired. A superseded decision gets a follow-on row referencing the prior; both stay. History is the point.
Cross-references. Every row lists downstream artifacts touched so that when a pivot lands, the artifact-update work is visible.
Seed pass complete: 14 pivots across 5 categories. v0.2 lands after the founder alignment meeting (2026-09-04) with any additional pivots the meeting produces. Next pivots likely to belong here: brand name (post-A1), Sindhu final title (post-A2), investor-lane priority (post-A7), data-source partnership marketplace structure, any advice-boundary reformulation.