Strategic Acquisition Brief · Confidential
Principal has stated the problem in its own words: roughly 60% of plan participants are unadvised, and Workplace Personal Investing set out to reach them — with about 160 salaried advisers. MaxiFi is Prof. Laurence Kotlikoff’s deterministic lifetime-planning engine that computes the provably correct plan for every household: sustainable spending, Social Security timing, Roth strategy, and the full tax code. The thesis Principal has already declared — run at machine speed.
Principal launched Workplace Personal Investing to reach the mass-affluent participants it has publicly said are being missed — roughly 60% of participants, unadvised — and staffed it with about 160 salaried advisers. The thesis is right, and it is Principal’s own. The constraint is arithmetic: millions of eligible households divided by 160 humans is a thesis running at human-planner speed.
On June 30, 2026, Principal expanded its income suite — LifeTime Income Builder Index target-date CITs allocating to guaranteed income from mid-career and targeting a stated payout at retirement, alongside third-party retirement-income TDF series. Converting savings into income raises, for every participant, the exact questions only a computed plan answers: how much income, from what, starting when, claiming when, converting what — and how much can I actually spend? A menu of income products without a computed recommendation is still a menu.
Meanwhile the appetite for the engine is already proven internally: a new CEO seated in 2025, and AI tools in daily use by thousands of Principal employees. What the stack is missing is not willingness — it is the one deterministic engine that turns the stated thesis into a shipped capability.
Principal did the hard part in public: it named the underserved population and committed to reaching it. What remains is the delivery math. An engine that computes the single correct plan per household is the only way the declared thesis serves the sixty percent — at the speed of software, not the speed of hiring.
A large language model is a horizontal capability. In any function where a wrong answer is catastrophic, no serious operator ships the raw model to the user — a purpose-built application layer sits on top of it, encoding the domain’s rules and holding the model to them, turning raw generation into an action that is correct, defensible, and safe to deploy. It is already how high-stakes AI gets built: Intuit runs a deterministic tax engine under TurboTax’s AI and won’t let the model compute the numbers; patient-facing healthcare AI runs inside a safety layer, not on a raw model; and no one boards a plane flown by an unverified black box.
Personal financial planning is exactly such a function, and getting it wrong is its own kind of disaster: the retiree who runs out of money at 82, the family under-insured by a million dollars. It is also precisely where a model, left alone, fails — because it reaches for the same rules of thumb the incumbents use, and in this domain approximation is not “close enough”; it is wrong, in ways that compound every year to the household’s detriment.
Built from the ground up over thirty years, MaxiFi is the proprietary workflow that computes the correct, auditable answer under the actual tax and benefit rules — the function users actually want performed. The model doesn’t have it. The user doesn’t have it. The incumbents approximate it — and, in this domain, approximating it means getting it wrong, to the household’s cost. Value accrues to whoever owns that trusted workflow: the model layer commoditizes, while durable value moves up to the layer that owns the user’s trust and performs the function.
MaxiFi is the financial-planning platform of Economic Security Planning, Inc., built over more than three decades by Professor Laurence Kotlikoff of Boston University. It uses consumption smoothing and dynamic programming to compute the single, mathematically optimal lifetime plan — solving simultaneously across Social Security strategy, Roth-conversion sequencing, withdrawal order, estate planning, and the full current tax code.
Goals-based tools answer “What is the chance you hit your number?” MaxiFi answers “What is the optimal path, and how much can I spend today without jeopardizing tomorrow?” It is not a better simulator. It is a different class of engine — deterministic, computed, and reproducible rather than sampled and probabilistic.
Prof. Laurence Kotlikoff — William Fairfield Warren Professor at Boston University; Harvard Ph.D.; former Senior Economist on the President’s Council of Economic Advisers; Fellow of the American Academy of Arts & Sciences and the Econometric Society; named by The Economist among the 25 most influential economists (2014). He intends to keep contributing to the product, help integrate, and stay on as spokesperson.
Taught by Nobel Laureate Robert Merton at MIT Sloan as an “outstanding science-based lifecycle and retirement management platform.” Merton uses MaxiFi as the reference engine in his MIT Sloan teaching. Featured in Bankrate’s “Best Financial Planning Software of 2025” roundup, cited best for near- and long-term tax planning and the decumulation phase. Economics that build on Nobel-laureate work.
30+ years of R&D in economic theory and dynamic programming — not scraped text, not a prompt-engineering layer. The kind of intellectual property a large language model cannot reconstruct by sampling tokens. Each year of refinement is time a competitor cannot buy back.
In every high-stakes AI deployment, a deterministic engine sits under the model — Intuit runs a tax engine under TurboTax’s AI and won’t let the model compute the numbers. Kotlikoff’s engine is that layer for lifetime planning: machine-speed, defensible answers under the adviser and under the AI.
Inside Principal, MaxiFi does one strategic thing: it converts the stated advice thesis from human-adviser speed to machine speed. MaxiFi computes each participant’s lifetime plan from plan and payroll data — sustainable spending, Social Security claiming, Roth sequencing, withdrawal order, and how much of a balance to convert to guaranteed income. The ~160 Workplace Personal Investing advisers inherit machine-prepared, deterministic answers instead of blank pages; the millions of participants no human will reach receive the computed answer directly, under Principal’s brand and duty of care.
And under the expanded income suite, MaxiFi is the decumulation brain: the computed, household-specific recommendation — how much income, from which product, starting when — that turns an income menu into an answer a participant, an adviser, and an examiner can each stand behind.
Every planning tool — and every AI trained on them — approximates the same heuristics: rules of thumb, replacement ratios, withdrawal shortcuts. As models improve, they converge on one another, and the industry mistakes that agreement for accuracy. A perfect mimic of an approximation is still an approximation — still wrong — and the error compounds a little further every year, to the participant’s detriment.
MaxiFi does not approximate. It computes — iteratively, multivariately, and simultaneously across taxes, benefits, longevity, and cash flow, year by year for a whole lifetime — arriving at the one optimal plan. The result is reproducible, with a full audit trail from data to answer. It is provable, not merely confident: the only answer that holds up when someone with an adverse interest checks the math. (The claim is about the computation being exact and inspectable — not a claim about predicting markets or investment returns.)
MaxiFi computes each participant’s lifetime plan from the plan and payroll data Principal already holds. Workplace Personal Investing advisers review, personalize, and own the relationship — the engine prepares, the human advises. For the population no human will reach, the computed answer ships directly, correct by construction.
Under the income suite, every option gets a computed, household-specific recommendation — how much to convert, starting when, claiming when — with a traceable calculation behind each number. The scaled offering is exam-defensible from day one, because defensibility is the design, not a retrofit.
SEC Reg Best Interest and FINRA suitability standards govern the substance of financial recommendations regardless of the interface that delivers them. Those rules are technology-neutral: the obligation follows the advice, not the medium. A firm that deploys an AI-assisted advice layer still owns the output that reaches its clients.
The 2026 exam environment has made this explicit. SEC 2026 exam priorities flag AI and technology risk: “if AI affects investor decision-making, it becomes an exam priority.” The FINRA 2026 Annual Regulatory Oversight Report dedicated a section to generative AI, naming the specific failure mode in client-facing agents:
“General-purpose AI agents may lack the necessary domain knowledge to effectively and consistently carry out complex and industry-specific tasks.”
“Complicated, multi-step agent reasoning tasks can make outcomes difficult to trace or explain, complicating auditability.”
That is a regulator describing an LLM-only financial-advice agent: wrong-prone and un-auditable. The direction is clear: AI is not a liability shield, and firms that deploy it in the advice channel own what it says. FINRA’s 2024 guidance (Reg Notice 24-09) already put firms on notice that these rules reach embedded vendor AI — “whether… developing Gen AI tools for their proprietary use or … leveraging the technology of a third party, including through embedded features in existing third-party products.” The 2026 cycle reaffirmed and deepened that posture.
For a firm scaling personalized advice across millions of plan participants, the practical question is not whether to comply but how to build so that compliance is not a retrofit. MaxiFi answers it at the design level. The engine computes the plan deterministically; the output is traceable and reproducible; the answer starts from “the most a household can safely spend with what it has” — sustainable by construction — rather than the aspirational “how much will you need” that manufactures the litigable number. A confident-wrong answer at consumer scale is not a compliance risk; it is a franchise risk.
Through late 2025, AI engines widely told users that the federal estate-tax exemption would “sunset” on January 1, 2026 — reverting from roughly $13.6M to $7M per person. The One Big Beautiful Bill Act, signed July 2025, instead permanently raised it to $15M per person. A confident, plausible, entirely wrong answer, delivered at scale to households making estate plans. MaxiFi computes against the current tax code; it cannot hallucinate a statutory change that did not occur.
The most useful third-party signal is also the most recent. On May 7, 2026, CBS MoneyWatch ran an identical retirement question — a 50-year-old single woman, retiring at 65 — through Claude, ChatGPT, and Perplexity. The verdicts diverged. Kotlikoff, quoted in the piece, noted that AI engines commonly mishandle Social Security timing by averaging longevity instead of using maximum life expectancy, and may “do more harm than good.” MIT finance professor Andrew Lo was also quoted, observing that AI systems have no “best-interest duty” analogous to a human advisor’s fiduciary obligation.
“Asked whether a 50-year-old single woman could retire at 65, Claude, ChatGPT, and Perplexity gave divergent answers. Kotlikoff: AI may ‘do more harm than good,’ mishandling Social Security timing and using average rather than maximum life expectancy. MIT’s Andrew Lo: AI lacks any ‘best-interest duty’ analogous to a fiduciary.”
The divergent-verdict story →AI engines widely told users the federal estate-tax exemption would “sunset” on January 1, 2026. The One Big Beautiful Bill Act (signed July 2025) instead permanently raised the exemption to $15M per person. Dollar-specific planning decisions were made on a wrong number — delivered with confidence, by every major engine, at consumer scale.
See the Kotlikoff estate test →Neither of these is an edge case. They are the structural failure mode of a probabilistic engine giving confident answers on a domain that requires deterministic computation. The divergent-verdict story is the named, neutral proof from the national press; the estate-tax error is the dated, dollar-specific example from the public record. Both point to the same gap — and the same engine that fills it.
Over a ten-week period in 2026, Larry published a six-post sequence on his Substack, Economics Matters — 137,000+ subscribers — running named frontier models against MaxiFi on real, dollar-specific household problems. Results are dated, reproducible, and verifiable. The stakes these posts name are the exact stakes Principal faces as it scales advice to the unadvised sixty percent.
“The AI said John and Jane can spend approximately $52,000 per year in discretionary spending. MaxiFi’s demonstrably correct answer — verifiable by inspecting its reports — is $63,382.”
Read the head-to-head →“Large language models are trained on text, not on solving optimal household financial problems. They have no internal model of taxes, Social Security, mortality risk, or lifetime budget constraints.”
Read the structural argument →“Claude understates John’s base plan’s final estate by 31 percent and his final plan’s final estate by 28 percent. On a re-prompt, Claude now says the final plan reduces John’s terminal estate by over $1 million.”
Read the estate test →“The median household leaves $182,370 of lifetime Social Security on the table. AI tells Jane a job change adds at most $35K in lifetime benefits when the right answer is $168K.”
Read the Social Security test →“I fed Claude all of John’s data. It concluded that John’s real sustainable discretionary spending was $167,000 per year — or 72.7 percent more than John can afford. If John were to spend at that level, he’d run out of money mid-retirement.”
Read the Roth test →Acquiring MaxiFi acquires the megaphone these pieces ship from — pointed, dated, and dollar-specific, at the exact audience Principal says it wants to reach. Larry intends to keep contributing to the product and to stay on as spokesperson, turning a category critic into Principal’s correctness narrator. The stated-gap story writes itself: the company that named the unadvised sixty percent acquires the engine built to serve them.
Principal named the gap in its own words — roughly 60% of participants unadvised — and launched Workplace Personal Investing to close it. MaxiFi is the engine that makes the stated thesis run: the single correct lifetime plan per household, computed at the speed of software. The story is already written; the asset is available now.
The June 30 expansion moves savings-to-income conversion to the center of the product line. Every income option implies a household-level recommendation — how much, from what, starting when. MaxiFi computes that recommendation, turning an income menu into a defensible answer — and validating the product rather than over-selling it.
You cannot serve millions of mass-affluent households with 160 advisers; you can with a deterministic engine under them. Advisers inherit machine-prepared plans; unreached participants get the computed answer directly. The stated ambition becomes a description of an actual system, not an aspiration.
MaxiFi’s answers are computed, reproducible, and auditable. They start from “the most a household can safely spend,” sustainable by construction — not the aspirational “how much will you need” that manufactures the litigable number. In the first FINRA exam cycle to treat GenAI accuracy as a standalone topic — and in a year when retirement-plan class actions are testing target-date design — defensible-by-construction is the control the environment is asking for.
The recordkeeper-to-wealth conversion race has more than one entrant, and there is one MaxiFi. Once placed elsewhere, Principal’s claim to a differentiated, defensible advice engine weakens permanently. Larry Kotlikoff intends to stay on with the acquirer in whatever capacity best serves the product — architect, spokesperson, advisor. A focused strategic process is underway.
Revenue — correctness converts unadvised participants into advised, retained, higher-value relationships at machine scale: attractant, adhesive, loyalty, continuity. Antidote & denial — owning the one computationally correct engine retires the scaled-advice liability overhang and denies the same asset to the rivals running the same race. Equity story — a differentiated, proprietary engine is what converts an AI budget into a defensible growth narrative a public company can take to its shareholders.
Larry built MaxiFi over thirty years for exactly the households Principal says it wants to reach. We are running a deliberately narrow process to place it where it serves the most everyday savers — and few channels reach the unadvised sixty percent the way the recordkeeper does.
MaxiFi is being offered through a focused strategic process. What is being acquired is the engine, its IP, and thirty years of R&D. For Principal the integration path is direct: the engine inside the employee-planning platform, machine-prepared plans handed to planners, a re-rated organic-growth trajectory. Founder continuity de-risks it: Larry Kotlikoff intends to stay on with the acquirer in whatever capacity best serves the product — architect, spokesperson, advisor.