A living log of program-management and product opportunities spotted while learning the tax business. This is the "so what would I DO about it" layer that sits on top of the reference files.
Status: Working — started 2026-07-09.
Nature of this doc: These are Victoria's hypotheses and analysis, NOT sourced fact and NOT claims that Intuit lacks these capabilities. Intuit has run the tax business for 40+ years and almost certainly has mature processes here. The value is the proactive lens and the questions — where could a known, recurring problem be made more proactive, more automated, more AI-augmented, or de-risked?
The insight: the yearly tax-code change isn't a surprise — it's a known, recurring, partially-predictable event. That makes it a candidate for a standing program that classifies each type of change by predictability and impact, then applies the right response — instead of treating every season as one undifferentiated scramble. (Source domain: file 03 §4 + file 07.)
The five change areas differ sharply in how predictable they are — which should drive different program responses:
| Change area | Predictability | Proactive program / product angle | Question to ask in the role |
|---|---|---|---|
| Inflation adjustments (brackets, standard deduction, contribution/phase-out limits) | High — formula-driven; IRS publishes indexed amounts on a schedule | Treat as data, not code — a parameterized config the software ingests, with an automated validation harness. Little/no hand-coding. | "How much of the annual inflation update is parameterized vs. manually implemented each year?" |
| New / expired credits & "extenders" | Medium — sunset dates are known, but extensions are politically timed and often late | Legislative horizon-scanning + a modular credit framework so adding/sunsetting a credit is config; scenario-plan the "will Congress extend?" branches | "How do you handle late-breaking extender decisions that land close to the season?" |
| New / revised forms & schedules | Medium-High — IRS releases draft forms before final | Track the IRS draft-forms calendar; start building against drafts; a forms-change intake pipeline mapped to the XML schema | "How early do you build against IRS draft forms vs. waiting for final versions?" |
| Big law changes (a major tax act) | Low, high-impact — episodic, sometimes retroactive or passed late in the year | The real risk area: a surge playbook, contingency capacity, a cross-functional war-room, and architecture that contains blast radius so one big change doesn't ripple everywhere | "What's the playbook when a major act passes in, say, December for the coming season?" |
| ATS / schema re-certification (IRS acceptance testing) | Scheduled, hard deadline | A certification critical path anchored to IRS ATS windows; automated conformance testing against the new schemas/business rules so it's continuous, not a season-end crunch | "Is IRS conformance testing automated/continuous, or is it a season-end push?" |
A "tax-year change management" program that does one disciplined thing: classify each incoming change by predictability × impact, and route it to the right response —
- Predictable/formulaic → automate (config/data pipelines, no hand-coding).
- Semi-predictable → horizon-scan + modularize (watch drafts and legislation; make changes config-shaped).
- Unpredictable/high-impact → surge-plan + contain (playbooks, buffer capacity, blast-radius architecture).
- Scheduled gates → critical-path + continuous testing.
This converts "annual scramble" into a portfolio of change types with tailored playbooks — a very PM-shaped reframe.
This is a natural place for AI to earn its keep, and it plays to a tech-forward PM's strengths:
- Detect — AI monitors IRS releases, draft forms, and legislation and flags what changed vs. last year.
- Draft — AI proposes the implementation delta (which rules/forms/limits change) for humans to verify.
- Test — AI accelerates conformance/ATS testing against the new schemas.
An "AI-augmented annual change pipeline" is a concrete, ownable program that connects domain + AI + execution — exactly the intersection this role is about.
The insight: execs already reach for AI tools to ask about business performance, but ungrounded AI answers wrong — while the ECC has the right data behind a filter-heavy UX that execs misread. Marrying them (ECC as system of record + a curated semantic/metric layer + AI that answers only from it, with receipts: as-of, tier, filter set) fixes both failure modes at once. Plus a pushed "first thing in AM" brief (page + podcast) and an external competitor/market digest.
Source domain: direct experience (Victoria led the ECC re-platform; saw the exec-AI friction firsthand). Full concept: intuit-role/exec-insights-concept.md · working synthetic demo: intuit-role/exec-insights-demo/.
Proactive angle: registry-first governance (top-20 exec metrics, owner-curated definitions/canonical cuts); push-brief-first sequencing (earns trust with zero NL risk); tier-aware answer confidence.
AI angle: the whole thing — but contained: registry-only answering, no free-form SQL, GenOS-native, provenance on every number.
Questions to ask: what's been attempted and why it stalled; ECC API/semantic-layer maturity; actual top-10 exec queries; GenOS integration path; who approves AI data access.
Caveats: prior attempts exist (details unknown); Intuit may already have this on a roadmap; all exec-query hypotheses unvalidated until inside. Prep-only artifact — raise as curiosity, not pitch.
## PO-N — <short title>
**The insight:** <the observation, and why it's a PM/product opportunity>
**Source domain:** <which reference file(s) this came from>
**Proactive angle:** <what a program/product could do to get ahead of it>
**AI angle (if any):** <where AI adds leverage>
**Questions to ask:** <curiosity-framed questions to pressure-test with insiders>
**Caveats:** <what Intuit likely already does; what's hypothesis vs. fact>
(Add PO-2, PO-3… as new opportunities surface while working through the KB.)