FelixDecision log
Field Notes

Decision log — discovery phase

The decision record from the April 2026 discovery cycle. Each entry carries a status line showing whether it stands, evolved, or was superseded. Decisions made since live in the Engineering design log.

2026-04-04 — Go/No-Go: GO (Lean Track)

Decision: Proceed with building a personal wealth manager app. Phase: Discovery Type: Product Reversible: Easily — lean investment, testable hypothesis, small downside.

Rationale:

  • Genuine white space validated: no product does comprehensive advisory-only for self-directed investors
  • Strong beachhead: tech workers with RSU-heavy comp at public companies
  • CSV-first approach makes initial build cheap ($100-$300K) and fast (4-5 months)
  • Incumbents structurally can't compete (AUM business model conflict)
  • Key risks (retention, X-ray value, WTP) are testable cheaply before scaling

Key risks accepted:

  1. Retention beyond first X-ray is unproven
  2. Advice-to-action gap may cap engagement
  3. Mezzi is building in the same space
  4. $10-$30/month pricing may limit to lifestyle business without expansion to full playbook

Next phase: Strategy — define positioning, pricing model, MVP feature set, and GTM approach.

Status (July 2026): stands. The Go held, and the accepted risks evolved as predicted — Mezzi moved fast (repriced to $299–$1,499/year and registered as an RIA within three months of this entry), and retention is now addressed in the design via the daily briefing / actions / verification loop, pending validation with real users. See the competitive analysis for the field as it stands.

2026-04-04 — Beachhead Segment: Tech Workers with Equity Comp at Public Companies

Decision: Lead with tech workers at public companies (FAANG, large-cap tech), age 28-34, $800K-$3M investable, RSU-heavy comp. Type: Strategic Reversible: Easily — can expand to adjacent segments once the product works.

Rationale: Scored highest on pain (5), WTP (5), complexity (5), reachability (5), size (4) across all segments analyzed. Quarterly RSU vesting creates recurring trigger. Cluster in reachable communities (Blind, Reddit, HN). Intra-company network effects possible.

Status (July 2026): dormant, not reversed. The current build is a friends-and-family deployment scoped to equities and ETFs, with no equity-comp features and a US-or-India jurisdiction model this decision never contemplated. The segment analysis remains the reference for GTM; whether the RSU wedge is still the opening move gets re-decided then.

2026-04-04 — Track Type: Lean

Decision: Lean track — build X-ray MVP, test with 20-50 users, validate retention before scaling. Type: Product Reversible: Easily — can upgrade to full track if validation succeeds.

Status (July 2026): evolved. Leaner still — a solo friends-and-family build for users numbering in the tens, at roughly $85–95/month total cost (see the cost model). The intent — validate retention before scaling — is unchanged.

2026-04-04 — MVP Scoping Direction: X-Ray + Analysis + TLH Flagging

Decision: MVP scoped as portfolio X-ray (cross-account view), portfolio analysis (allocation, overlap, fees, concentration, asset location), and TLH opportunity flagging at position level. Two options remain on the table (scorecard vs. X-ray focus) — final call in Strategy. Type: Product Reversible: Easily.

Status (July 2026): superseded. Neither option was built as scoped. The deferred "final call" resolved into a bigger shape: the full advisory loop — nightly refresh, deterministic fact catalog, daily AI briefing with structured actions, a verification loop, news, and chat. The X-ray analytics survive inside the analytics engine; the product shape is in Surfaces.

2026-04-04 — Data Ingestion: CSV-First

Decision: Start with CSV/PDF upload for data ingestion. Aggregator (Plaid) and direct broker APIs deferred to later versions. Type: Technical Reversible: Easily — can add aggregator or APIs at any time.

Rationale: CSV is free (no aggregator cost), gives best data quality (brokerage exports are source of truth), simplest to build, and sufficient for a point-in-time X-ray MVP. Aggregator adds value when real-time monitoring becomes the priority.

Status (July 2026): refined. CSV import is still build phase 1, but the rail decision this entry deferred is now made: SnapTrade (read-only, brokerage-native) for the US and the CDSL/NSDL CAS backbone for India, with nightly re-pricing making the dashboard daily-fresh without daily account pulls. Plaid remains deferred. See data connectivity.

2026-04-04 — Pricing Direction: $10-30/month for MVP

Decision: Target $10-$30/month subscription for the X-ray MVP. Expansion to $50-$150/month as full advisor playbook is built. Type: Product Reversible: Easily.

Rationale: Comparable to premium finance apps (Monarch $15/month, Copilot $15/month). With CSV-first, unit economics work at this range. Higher pricing appropriate once feature set expands.

Status (July 2026): deferred. No fees at the friends-and-family level — which is also what keeps Felix outside the heavy advisory regimes for now. When pricing returns, note that Mezzi has since validated the higher band this research identified, repricing to $299–$1,499/year.