Strategic Acquisition Brief · Confidential
In one summer MassMutual published research finding that “the demand for credible, expert financial guidance has never been higher,” placed in the top ten of its industry’s AI index for a second year, passed $2 billion of advisory annuity sales, and heard its chief executive say the country has not prepared for hundred-year lives. It had already called planning the zipper between protection and wealth and built a Private Wealth division to put more advanced planning in front of its 6,000-plus affiliated advisors (as reported at the division’s February 9, 2026 launch). The plan is the product. What the stack does not yet contain is the instrument that computes, for each household, the most it can safely spend, whether it needs to take investment risk at all, and the protection the plan requires. MaxiFi is that instrument: the only commercial engine built on the economics of lifetime planning, faithful to the science and to the law, and warrantable because it is.
On July 18, 2026 MassMutual published its 2026 Financial Habits research: 62 percent of Americans at least sometimes avoid important financial decisions from anxiety or overwhelm; 82 percent say an advisor would help; 81 percent say today’s complexity makes expert advice more valuable; 74 percent say there is too much conflicting advice online; only 34 percent sought advice from a traditional advisor in the past year. Vaughn Bowman, Head of Wealth Management: “Our research shows that the demand for credible, expert financial guidance has never been higher, and it’s on all of us in the industry to help people cut through the noise and make informed decisions with confidence.” In May he had set the frame: “Financial planning is the connective tissue—the ‘zipper’—that brings protection and wealth together,” and “the fastest-growing advisors will be part of well-structured teams using agentic AI to prepare and assist with client interactions.”
On July 14 Roger Crandall told Fortune: “We’re heading to a society where people live for 100 years, and we haven’t really prepared for that.” On June 23 MassMutual placed tenth on the 2026 Evident AI Index for Insurance and third in innovation, the top mutual for a second year. On August 11 MassMutual Ascend passed $2 billion of lifetime advisory annuity sales through 1,700-plus independent advisors, annuities that, in its words, “help fill structural gaps in client portfolios.” On August 24 the head of its private-wealth consultants said clients “are not looking for another pie chart”; they need someone to “break down the complexities and provide options and considerations that include both upsides and downsides.” And on September 2 MassMutual agreed to sell control of Flourish, the wealthtech platform it built to $8 billion of custody, to an investor whose business is scaling wealthtech, keeping a significant stake: the firm owns what is strategic and lets others scale what is not.
A plan that holds to 100 rather than to life expectancy, a life-insurance need sized from the household’s own lifetime problem rather than a replacement ratio, an annuity placed where the safe base requires it, a Roth conversion timed against IRMAA and the survivor’s bracket: each is arithmetic under the law, and none of it is in the workspace, the questionnaire or the assistant. An assistant that prepares the conversation must call something that computes the answer. This is the layer that does.
Conventional planning gets the first thing wrong: it never asks whether a household needs to take risk at all. It sets a spending target, usually too high, and adds equity to raise the probability of hitting it. The economics of lifetime planning begins elsewhere: with the most a household can sustainably spend with no investment risk, computed jointly across its whole life under the taxes and benefits the law actually imposes, with spending free to adjust, and only then treats risk as an option, evaluated against the household’s own aversion to loss. For a household that pays for a guarantee rather than a probability, that is not a rounding error. It is the answer, in the household’s own terms, and it is the science’s own case for the protection and the guaranteed income a mutual sells.
Laurence Kotlikoff’s August 14, 2026 case shows what the premise is worth to an illustrative cautious household: a hypothetical 62-year-old retiree with $2 million and Social Security.
| Strategy | Sustainable discretionary spending | Versus the safe base, in the household’s own terms |
|---|---|---|
| Safety-first (TIPS-only base) | $83,451 / yr to age 100 | The base case: no investment risk at all. |
| 80/20 stocks and bonds | Far higher expected, much greater dispersion | 22% worse if the household is highly risk-averse; 22% better if it is risk-tolerant. |
| 20/80 stocks and bonds | Modestly higher expected | 9% worse if the household is highly risk-averse; 6% better if it is risk-tolerant. |
| Claim Social Security at 70 and time Roth conversions | + $8,608 / yr | Risk-free. Computed, not guessed — the part of the answer no glide path produces. |
Figures are Professor Kotlikoff’s, as of September 2026, for his illustrative household’s facts and assumptions, based on his August 14, 2026 article; they are not generalized. The plan depends on the person, and conventional planning never asks.
A mutual whose policyowners buy guarantees, whose chief executive plans for hundred-year lives, and whose head of wealth management calls planning the zipper between protection and wealth is the firm for which “does this household need to take risk at all” is the native question. An engine that begins from safety is the only kind that can answer it, household by household; it sizes the protection and places the annuity from the same computation; and it is the only kind whose answer a carrier can stand behind.
A large language model is a horizontal capability. Wherever a wrong answer is catastrophic, a purpose-built layer sits on top of it: Intuit runs a deterministic tax engine under TurboTax’s AI and does not let the model guess the numbers. MaxiFi is that layer for lifetime financial planning. It is not an application MassMutual would operate; it is a computation that the advisor’s workspace, the private-wealth consultant and the assistant MassMutual is building all call. From the client data the firm already holds, it produces the safe base, the claiming and conversion sequence that fits that household, the survivor’s plan, the life-insurance need year by year, and the risk the household may choose to take on top. The client’s experience does not change. The provenance of the number on it does.
MassMutual has shown this year how it thinks about software: cap the AI relationships at twelve months to keep optionality, and bring in an investor to scale a platform the firm does not need to run alone. Both are right for every layer that will change. MaxiFi is not a platform and not an application; it is intellectual property, the reference layer beneath any interface, as actuarial tables sit beneath any policy, and it is the one layer that must not change under the assistant. A rented planner cannot be warranted, because its spending target comes from outside the model and there is no correct answer to warrant. An owned engine that is faithful to the economics and computes under current law can be, and Larry Kotlikoff stays with it as architect. Owned inside MassMutual, that is a claim no other carrier can make.
The life-cycle economics of saving, spending and risk has been taught in graduate finance for seventy years. MaxiFi is the only commercial engine built on it, and it is faithful to the science in two layers. The economics: safety-first, spending endogenous, the lifetime solved jointly, risk evaluated as an option against the household’s own risk aversion. The law: federal and forty-two state tax codes, Social Security, Medicare and IRMAA, required minimum distributions, Roth conversions, pensions, annuities and survivor benefits, encoded and maintained for thirty years, deterministic in the sense that matters — the same facts produce the same auditable answer.
When a technology team says “we could build this,” the answer is not the algorithm. Dynamic programming is taught. The answer is the classes of failure in which a rebuild silently ships a wrong number:
Patent-winning. Named to Bankrate’s best financial planning software of 2025. Used by Robert Merton to teach financial planning at MIT Sloan. Built over thirty years by Boston University economist Laurence Kotlikoff, who intends to stay with the acquirer as architect and spokesperson. An active base of households is already planning on the method; figures in the data room.
For a client’s facts and assumptions, the plan MaxiFi produces is the plan the economics prescribes, computed under current law. A computational error is objectively decidable: rerun the engine and check. That is why a bounded accuracy guarantee with a stated exclusive remedy is insurable for an engine that owns its computation, and is not offerable by a planner whose target is set outside the model. Market returns, longevity, future law and the client’s own assumptions are never guaranteed. The guarantee covers the arithmetic and the fidelity, not the life.
The setting is specific. MassMutual’s affiliated advisors practice through a FINRA member and SEC-registered adviser, and FINRA’s Regulatory Notice 24-09 (June 2024) says the existing rules apply in full to advice produced with generative AI; FINRA’s 2026 oversight report says controls “should address hallucinations,” the SEC’s 2026 examination priorities ask whether “algorithms produce advice or recommendations consistent with investors’ investment profiles,” and the NAIC’s AI bulletin, adopted by 23 states, asks insurers for “documentation of model origins and standards for explainability” in the third-party AI they use. A wrong claiming age, conversion, protection amount or survivor plan from an assistant in front of a mutual’s policyowners is the same wrong answer for every household that asks; the cost is restitution to the households that relied on it, the examination that follows, and the erosion of a 175-year reputation for keeping promises. The antidote is a reference answer: the assistant’s numbers are the engine’s numbers, rerun and checked against the law in force on the plan date.
It converts “credible guidance” into advice a carrier can defend as the economics: in front of a client, a recruit choosing a firm, an examiner and a plaintiff. The protection recommendation becomes a computed need rather than a quota’d one; the annuity becomes the base case the science prescribes rather than a product the plan was fitted to. We are not selling an insurance policy. The insurance is included.
Independent tests in 2026 established that the market cannot check its own arithmetic. Larry Kotlikoff’s Economics Matters Substack has run named head-to-heads against frontier engines on dollar-specific household problems; the variance across engines on identical prompts is the proof that correctness cannot come from the model layer.
How will you validate that any AI is correct? You will need to run MaxiFi to know. Correct means faithful to the economics; the instrument that can certify the AI is the faithful implementation of the science, computed under current law, today and after any attempted replication.
MassMutual’s trajectory is planning-led wealth across a mutual’s advisor force, with agentic assistance, for households that will live to 100, at a scale of $312 billion of client assets and $1.1 trillion of protection. MaxiFi is not an addition to that trajectory. It is what completes it: the computed, warrantable answer to the question every one of those households eventually asks, and the science’s own case for the guarantees the firm sells.
Planning brings protection and wealth together, in MassMutual’s words. The engine is what makes that a number: the safe base, the claiming and conversion sequence, the survivor’s plan and the protection need, solved jointly in the workspace the advisors already use. Provable in the first quarter of ownership.
Safety-first planning is the case for guaranteed income and for a computed life-insurance need, made by the economics rather than by the product desk. The recommendation is defensible because it is the solution to the household’s own problem, and a frontier model asked the same question is off by millions.
A Private Wealth division promises more advanced planning; an advisor choosing a firm asks what the planning is worth. A warranted plan, computed under forty-two state codes, is the claim that answers both, and the one no other carrier’s recruiter can match.
An adviser is judged against the economics, not against any vendor’s product. The firm whose advice is faithful to the science has the defense; the one whose tool is not has the exposure. Owned inside MassMutual, the engine is denied to every other carrier; the mutual that keeps promises becomes the carrier that can warrant the plan.
The workspace holds the plan; the assistant prepares the conversation; neither decides the household’s number, and the household’s number is what the policyowner’s trust is about. A rented planner cannot be warranted. An owned engine can, and there is exactly one MaxiFi.
MaxiFi is offered for acquisition: the engine, its intellectual property, thirty years of encoded law, and Larry Kotlikoff’s continued involvement as architect and spokesperson. The next step is a 30-minute orientation: one real, anonymized household, our machine, in the room — the safe base case first, then the risk question — while a frontier model is asked to match it. Nothing is deployed and nothing is left behind; evidence deepens with commitment, and the full case is provable in an acquirer’s first quarter of ownership. First conversations with strategics are underway; we expect to narrow the field in early November.
Or write directly: Commerce@kaneco.com · subject “MaxiFi — MassMutual orientation.”
Thirty minutes: the safe base case for a real, anonymized household first, then the risk question, then a frontier model asked to match the result. Choose a window and book it directly; or email with another time.
Commerce@kaneco.com