The math layer Anthropic's FS agents will need.

Algorithmic fiduciary by construction.

MaxiFi's objective function is the household's lifetime consumption. It cannot recommend a product, an AUM strategy, or an asset gathering action that does not serve the user — because none of those quantities appear in its loss function. This is constitutional alignment, transposed from AI safety into finance.

Math an LLM structurally can't do.

30 years of dynamic programming over consumption, taxes, claiming choices, and longevity risk. Returns answers that are deterministic, replicable, and auditable. Not approximations. Not pattern-matched intuitions. Actual answers.

Tested directly. Fed an identical household scenario, Gemini under-estimated sustainable spending by 13% across forty years — effectively denying ~18% of the household's assets. With a small variation, ChatGPT over-estimated first-year safe spending by nearly 2×. Both LLMs reported high confidence; one stated it was "highly confident numerically." Both were wrong.

And 13% is the floor, not the ceiling. The error compounds forward through time and laterally through follow-up questions — Roth conversion timing, claiming age, drawdown order — each of which is now answered against a baseline that is already wrong. By the time a household has worked through a normal sequence of planning questions with an LLM, what it has is not an answer that is 13% off. It is a structurally wrong plan. (Published comparisons, May 2026.)

A trade-secret moat that cannot be trained for.

The engine cannot be replicated by a larger training run. It is the product of three decades of economist-led structured problem solving. A competitor starting today is starting thirty years behind.

Regulatory cover.

SEC Reg BI, DOL fiduciary, FINRA AI supervision guidance, and EU AI Act treatment of financial advice as "high-risk" all converge toward demanding auditable reasoning under any AI-generated financial advice. MaxiFi is structurally compliant. LLM-only stacks are structurally not.

The window is open now — on two surfaces.

Anthropic committed publicly to FS leadership on May 5. Ten days later, OpenAI launched ChatGPT Personal Finance — Plaid-integrated across more than twelve thousand U.S. financial institutions, built on the team it acquired with personal-finance startup Hiro. The FS conversation now runs in parallel across the professional surface and the consumer surface. MaxiFi is the deterministic engine underneath both.

The professional surface — already in use.

The May 5 pro-tier agents — pitch builders, KYC, audit, household-advisory copilots — are sold into financial institutions, RIAs, and large wealth managers. MaxiFi is in that channel today via its Pro subscription: fee-only advisors use it as top-of-funnel lure to win prospects, and as the recurring lifecycle math that keeps client relationships annual. The May 5 slate gets a real upgrade when MaxiFi sits beneath it as substrate.

The consumer surface — being claimed in real time.

OpenAI's May 15 launch has put the consumer-finance surface in motion. Whatever Anthropic mounts in response will need a deterministic engine that returns correct answers under regulatory scrutiny. The math under that surface is also MaxiFi. One asset. Two surfaces.

Competitive denial.

If MaxiFi lands at OpenAI, Google DeepMind, Meta, or a large fintech, Anthropic's claim to FS leadership weakens permanently — because the most defensible single piece of FS infrastructure is now owned by a competitor. The denial case sits alongside the offensive one.

The Kotlikoff megaphone.

Acquiring MaxiFi acquires Larry Kotlikoff — Boston University economist, author of Money Magic, Spend Til The End, The Coming Generational Storm; longtime PBS, Forbes, and Yahoo Finance columnist; 2016 presidential write-in candidate. He is a credibility distribution channel pointed directly at the FS narrative Anthropic is building — across both surfaces. There is no comparable analog in AI M&A.