Probate & estate settlement

Every death hands one person — usually a grieving relative with no legal training — a months-long, multi-party paper chase: open a case in the county probate court, track down every account across dozens of institutions that expose no API, notice creditors, file the final tax returns, and distribute what’s left to heirs. It runs an average of ~16 months and ~570 hours of executor effort per estate (EstateExec), on near-universal, non-cyclical volume. The agent opportunity is to own the discovery → forms → institutional-chase → accounting loop end to end, while the executor keeps the signature and the court keeps the gate.

Vitals: market: $84T+ wealth transfer through 2045 (Cerulli) · ~2.6M US probate cases/yr · buyer: executors / estate attorneys / fiduciaries / banks · model: per-estate fee or % of estate · whitespace: ★★

Market context — size, money flow, who profits today

• The wave: Cerulli projects $84.4T transferred through 2045 — $72.6T to heirs, $11.9T to charity (Cerulli). Most of it passes through, or just around, the estate-settlement process.
• The volume: ~2.6M probate cases are filed annually in US state courts (NCSC, 2023, reported via secondary source); academic estimates put 15–40% of all deaths into probate (same), against ~3M+ US deaths a year.
• Who pays today: fees flow to estate attorneys and CPAs (the average estate spends $12.4K on legal + accounting alone; executors take ~$18K in compensation — EstateExec), and to corporate fiduciaries (bank trust departments) on larger estates.
• Tax is rarely the issue: only estates >$13M owe federal estate tax — <0.1% of estates (EstateExec). The work is administrative, not tax-planning; the pain is process, not liability.

The mess

• Asset discovery has no API. The executor reconstructs the deceased’s financial life by mail and phone — every bank, brokerage, 401(k), IRA, pension, life policy, and utility — through hold queues and notarized-letter requirements. Alix, an AI-native entrant, markets exactly this gap: it “did a wonderful job of finding different assets that otherwise would have gone unnoticed” (Alix).
• Court forms are county-specific with no standard. Filing the petition, getting Letters issued, and filing the inventory and final accounting vary by jurisdiction — which is why incumbents sell “state-specific guidance” as the product (EstateExec) rather than a generic form-filler.
• The clock is long and the effort is brutal. ~16 months to settle, ~570 hours of executor work, and 80% finish within 18 months (EstateExec) — a multi-month, stateful, deadline-bound workflow (creditor-claim windows, tax deadlines) running on spreadsheets and a filing cabinet.
• Fees scale with the gross estate, not the work. In statutory-fee states the attorney and the personal representative each draw a percentage of the gross estate — in California, 4% of the first $100K, 3% of the next, 2% of the next $800K, so a $1M estate pays roughly $23K each, ~$46K total (Cal. Probate Code schedule), regardless of how mechanical the administration was.

Why now

The demographic wave is once-in-history — the largest wealth transfer ever recorded is moving through an artisanal, paper-bound process built for a fraction of the volume. What changed on the supply side is capability: an agent can now read wildly varied institutional documents (statements, beneficiary forms, deeds, court notices), reason about what’s missing, and drive a long-running, multi-party workflow with deadlines and dependencies — the two things estate settlement is made of.

Why it stayed unsolved is structural and is exactly why it’s defensible: the process is fragmented county-by-county with no APIs, it’s emotionally loaded, and it’s low-frequency per person — almost nobody settles two estates, so no consumer ever climbs the learning curve, and the knowledge stays locked in solo attorneys’ heads. The arcaneness is the moat (see Opportunities thesis).

The money

Signal	Figure	Basis
Generational wealth transfer	$84.4T through 2045 ($72.6T to heirs)	Cerulli (cer) — VERIFIED
Probate volume	~2.6M cases/yr in US state courts	NCSC 2023, reported (pcb) — INFERRED (secondary)
Time to settle	~16 months; 80% within 18	EstateExec study (ee) — VERIFIED
Executor effort	~570 hours per estate	EstateExec (ee) — VERIFIED
Legal + accounting spend	~$12.4K avg per estate	EstateExec (ee) — VERIFIED
Total settlement cost	4–7% of gross estate (national)	Multiple firms (caf) — VERIFIED
CA statutory fee	~$46K on a $1M estate (attorney + PR, on gross)	Cal. Probate Code schedule (caf) — VERIFIED

The business model is fee-for-administration, not clean contingency

Unlike property-tax appeals or subrogation, there’s no pool of “found money” to bill against — value capture is a per-estate fee (Alix ≈1% of estate / $9K min, reported; EstateExec a flat $199, ee). Debt-negotiation savings are an upside, not the core meter.

How it works today

The executor is the single manual hub. Everything routes through one untrained person coordinating the court, every financial institution, creditors, the IRS, and the heirs — sequentially, on paper and phone, for over a year.

Mermaid source

flowchart LR
  classDef human fill:#fdecec,stroke:#e0564f,stroke-width:1.5px,color:#0f172a;
  classDef pro fill:#eef0fe,stroke:#6366f1,stroke-width:1.5px,color:#0f172a;
  classDef court fill:#e8f1fd,stroke:#2563eb,stroke-width:1.5px,color:#0f172a;
  classDef ext fill:#fff7ed,stroke:#d97706,stroke-width:1.5px,color:#0f172a;

  Exec("Executor + family<br/>the manual hub<br/>~570 hrs over ~16 months"):::human

  Atty("Attorney + CPA<br/>paid a % of the<br/>gross estate"):::pro

  subgraph Court["Probate court · county-specific forms, no standard"]
    direction TB
    Pet("Petition + original will"):::court
    Let("Letters issued<br/>executor empowered"):::court
    Acct("Inventory + final<br/>accounting filed"):::court
  end

  subgraph Inst["Asset discovery · no API · mail + phone + hold queues"]
    direction TB
    Banks("Banks · brokerages"):::ext
    Ret("401k · IRA · pensions"):::ext
    Ins("Life insurance · annuities"):::ext
    Prop("Home · vehicles · belongings"):::ext
  end

  Cred("Creditors<br/>statutory notice +<br/>claims window"):::ext
  Tax("IRS + state<br/>final 1040 · estate returns"):::ext
  Heirs("Heirs / beneficiaries<br/>44% see family conflict"):::human

  Exec --> Atty
  Exec --> Pet
  Pet --> Let --> Acct
  Exec --> Banks
  Exec --> Ret
  Exec --> Ins
  Exec --> Prop
  Exec --> Cred
  Exec --> Tax
  Acct --> Heirs
  Exec --> Heirs

Where an agent fits

The agent owns the labor and the state; the human owns the judgment and the signature. It runs the discovery sweep (read statements, trace dormant accounts), generates the county-specific filings and creditor notices, works the no-API institutional surfaces over months, keeps a running inventory and estate accounting, and assembles the tax returns — surviving the multi-month timeline with its deadlines intact. The executor gates anything legally binding: e-signing court filings and approving the final distribution. Every action and document is logged, which doubles as the court accounting and the defense against the 44% odds of family conflict.

This is squarely the playbook’s territory: an agent that acts, not suggests; a durable, long-running workflow that can’t lose state across 16 months; integration with systems that have no API (portals, mail, phone trees); and per-jurisdiction rules encoded as a domain layer rather than hand-held per county.

Mermaid source

flowchart LR
  classDef human fill:#fdecec,stroke:#e0564f,stroke-width:1.5px,color:#0f172a;
  classDef agent fill:#eafbf1,stroke:#16a34a,stroke-width:1.5px,color:#0f172a;
  classDef court fill:#e8f1fd,stroke:#2563eb,stroke-width:1.5px,color:#0f172a;
  classDef ext fill:#fff7ed,stroke:#d97706,stroke-width:1.5px,color:#0f172a;
  classDef store fill:#eef0fe,stroke:#6366f1,stroke-width:1.5px,color:#0f172a;

  Exec("Executor<br/>gates signatures +<br/>final distribution"):::human

  Agent("Estate-settlement agent<br/>durable, multi-month workflow<br/>acts, not just suggests"):::agent

  subgraph Owns["What the agent owns"]
    direction TB
    Disc("Asset discovery<br/>read statements, trace accounts"):::agent
    Forms("Generate county-specific<br/>forms + creditor notices"):::agent
    Chase("Chase institutions<br/>portals · mail · phone trees"):::agent
    Acct("Running inventory<br/>+ estate accounting"):::agent
    TaxP("Assemble tax returns"):::agent
  end

  Inst("Banks · brokerages · IRS · creditors<br/>no-API, paper + portal surfaces"):::ext
  Court("Probate court"):::court
  Audit[("Audit trail<br/>every action + document logged")]:::store

  Sign{{"Executor e-signs filings"}}:::human
  Dist{{"Court-approved distribution"}}:::human

  Exec --> Agent --> Owns
  Owns -. "operates over" .-> Inst
  Owns --> Sign --> Court
  Court --> Dist
  Agent -. "logs to" .-> Audit

Whitespace & incumbents

The seed thesis called this a near-empty blue ocean. The deep dive corrects that: the AI-native wave arrived in 2025, so this is now a real-but-shallow field, not an empty one — hence ★★, not ★★★. The competitive map sorts into three bands:

• DIY executor software — EstateExec (“the #1 Software for Estate Executors,” a flat $199/estate, “TurboTax for executors,” with “state-specific guidance with AI software”) and Atticus (“the #1 DIY software for executors,” pairing software with an in-house expert team). These guide the executor; they don’t do the work.
• Service + platform — ClearEstate (“our platform and experts will handle the entire settlement process”) and Alix, the one to watch: an AI-native settlement service with human Settlement Specialists, $20M Series A in July 2025 (Acrew, Charles Schwab, Edward Jones Ventures, reported), proven “across estates from $20K to $20M,” whose “AI agents… [scan] documents, [extract] data, [pre-populate] legal forms” (Alix).
• B2B fiduciary tooling — Estateably (“the first cloud-based platform for estate administrators,” fiduciary accounting and “intelligent automation”), sold to the professionals rather than the family.

The gap is autonomy depth, not absence

Most incumbents are checklist/guidance software or human-led services with AI assist; the durable, act-not-suggest agent that runs the full institutional chase with the audit trail as the artifact is still nascent. The bet is now execution depth against Alix-class competition — no longer a land grab on empty space.

Hard problems

Problem	Why it’s hard here	Signal	Likely approach (speculative)
Asset discovery with no API	Accounts are scattered across institutions that expose no programmatic access; the agent works mail, portals, and phone trees, and must reason about what’s missing	”found different assets that otherwise would have gone unnoticed” (Alix)	Probably browser/RPA + document-extraction agents over institutional portals, plus heuristics that flag implied-but-unfound accounts (1099s without a matching account, etc.) — see integrating systems without APIs
Per-jurisdiction court rules	Forms, deadlines, and accounting formats vary by county/state with no standard; getting one wrong restarts a clock	Incumbents sell “state-specific guidance” as the product (ee)	Likely a per-jurisdiction rules/templates layer (a DSL of forms + deadlines) rather than prompt-per-county — see encoding domain rules
16-month durable state	The workflow spans deadlines and dependencies across >a year; losing state or missing a creditor window is unrecoverable	~16 mo avg; 80% within 18 (ee)	Probably a durable execution engine (Temporal-style) with timers for statutory windows — see durable long-running workflows
Trust + the conflict surface	Money + grief + 44% family-conflict rate makes every action contestable; errors aren’t just costly, they’re litigated	”Over 44% of respondents had experienced or were aware of … family conflict” (ee)	Likely human-gated signatures + a complete, court-grade audit log of every action and document as the defensible artifact

Sources

• Cerulli — $84.4T wealth transfer through 2045
• EstateExec statistics — settlement time, effort, cost, conflict, estate-size data (vendor study, n>1,200)
• Alix · Series A coverage · pricing comparison — AI-native estate-settlement entrant
• Atticus · ClearEstate · Estateably — incumbent software/services
• California statutory probate-fee schedule — % of gross estate
• Probate statistics (NCSC, reported) — case volume, % of deaths

Reconstructed from public sources; claims are tier-labeled (VERIFIED / INFERRED / SPECULATIVE) — see how to read the tiers. Supporting quotes live in this repo’s evidence map (evidence/opp-probate-estate-settlement-evidence-map.md).