A retirement income company for small business owners, powered by AI automation. We acquire the businesses boomers built, operate them with agents and dedicated account managers, and pay the original owners a contractual, inheritable income through IRC 453 installment sales.
ExecuSystems is a B2B procurement intermediary my dad started in the early 1990s. His anchor client is Boot Barn — he's been working with them for 15+ years, sourcing everything from clothes racks to store signage for new store openings as Boot Barn expands. He negotiates with vendors, manages logistics, handles exceptions, and maintains relationships. He even built custom software on top of Zebra's catalog system — a portal with approval chains and workflow automation. It generates $437K in revenue annually at 37.9% gross margins, plus another $90K gross ($50K net) from a side Amazon arbitrage operation. It's pretty hands-off — he mostly talks to Boot Barn's VP of Operations. And it cannot survive without him.
Not because the business model is broken. Because the business is him. When Boot Barn sends a PO for clothes racks for a new store opening with a tight delivery window, my dad knows which vendor has stock, what substitution the client will accept if the preferred model is backordered, and what pricing exception applies based on this specific client's history. That knowledge — built from 15+ years with Boot Barn and decades before that with Sport Chalet (a previous large client that went bankrupt) — lives entirely in his head. There's no employee to promote. No buyer interested in acquiring a one-person operation. No succession plan.
So when he retires in the next 2-3 years, Boot Barn will find another vendor, and ExecuSystems will cease to exist. A business that generated $10M+ in cumulative profit over three decades will produce exactly zero dollars for his retirement because it cannot operate without him physically answering emails.
The realization: ExecuSystems isn't unique. It's the template for ~2.3 million small businesses in the United States. Profitable, cash-flowing operations run by solo owners or tiny teams where the owner's knowledge IS the asset and their retirement IS the end of the business. B2B procurement, insurance agencies, freight brokers, property management, bookkeeping practices — different industries, identical structure.
Every day, 10,000 baby boomers retire. The SBA estimates 70% of small businesses will change hands in the next decade. The majority have no succession plan. These aren't failing businesses — they're information businesses disguised as services where the owner's coordination knowledge cannot be transferred because it was never captured as data.
Private equity won't touch them (diligence costs more than the deal at this scale). Traditional buyers want businesses that can operate without the owner (these can't). Brokers get sellers 0.7-0.9x fair value after taxes and fees (then sellers figure out how to not outlive the money). There's a massive, unserved market for businesses worth $500K-$2M where the existing exit options are uniformly terrible.
That's what The Owners Endowment is. A retirement income company for business owners like my dad. We acquire the businesses boomers built, operate them with AI agents and dedicated account managers who inherit the owner's knowledge, and pay the original owners a contractual, inheritable income derived from the business they spent their life building. Not a lump-sum buyout. Not seller financing with default risk. A 15-year IRC 453 installment sale backed by a diversified portfolio of businesses, plus growth dividend participation as the portfolio scales.
And it turns out ExecuSystems isn't just the inspiration — it's the proof of concept. We have a free first deal (family business), an existing anchor client relationship to validate retention assumptions, and deep domain knowledge of the category. The entire 2026 plan is: transition ExecuSystems while still employed at Canva, validate that Boot Barn stays through an AI-powered ownership transition, and use that validation to acquire 3-5 more businesses with personal capital before deciding whether to go full-time.
Zero career risk. Massive asymmetry. And if it works, the model scales to every category with the same pattern.
These businesses were uninvestable because they couldn't operate without their owner. The owner's daily work — answering vendor emails, managing procurement exceptions, renewing client contracts, coordinating logistics — was irreplaceable human judgment.
That changed. AI agents can now autonomously complete tasks that previously required hours of skilled human coordination. According to METR's evaluations, the time horizon of reliable autonomous AI work has been doubling every 4-7 months, and frontier models can already handle multi-hour complex workflows.
The key insight: Most of these businesses are information businesses disguised as services. The owner's value isn't physical labor — it's coordination knowledge. "When Boot Barn sends this type of PO, look up the item in Zebra's catalog, apply our margin, and confirm delivery timeline." That's an agent workflow, not a human one.
But the owner's real value isn't just their workflows — it's 20-30 years of accumulated decision knowledge. When Boot Barn sends an unusual PO, the owner knows which vendor to call, what substitution the client will accept, and what pricing exception applies — based on hundreds of similar situations they've navigated before. That knowledge has never been captured as data. It lives in one person's head and dies when they retire. The Agent OS captures it for the first time.
This creates a new asset class: small services businesses that can be acquired cheaply and operated at dramatically higher margins through AI automation.
Critically, this is not full automation. Clients want to talk to humans for sales, relationship management, and issue resolution. But for everything behind the client relationship — order processing, invoicing, vendor coordination, inventory tracking, compliance, scheduling — agents are genuinely better: faster, fewer errors, available 24/7, no training ramp. The operating model is a split architecture: dedicated human account managers handle the client relationship layer while agents handle 60-70% of the operational work underneath. One account manager covers 8-10 businesses when the back-office is automated. That's the margin unlock.
Every seller receives two layers of compensation structured to avoid securities classification while providing tax advantages over traditional sales:
Sellers choose from three installment structures, all NPV-matched at 8% discount rate to the same fair market value. Note: The $600K example below is illustrative — deal sizes vary based on each business's revenue and profitability. Each structure includes a 2% annual escalator for inflation protection:
| Structure | Duration | Initial Payment | Escalator | Total Paid | Best For |
|---|---|---|---|---|---|
| Accelerated | 7 years | $85,000/year | None | $595K | Sellers needing faster liquidity |
| Standard (default) | 10 years | $60,000/year | +2%/year | $657K | Balanced, inflation-protected |
| Legacy | 15 years | $43,000/year | +2%/year | $627K | Maximum duration, inheritable |
Tax advantage: IRC Section 453 installment sales defer capital gains over the payment period instead of triggering a lump-sum tax event. A seller receiving $600K upfront pays ~$150K in federal capital gains immediately. Under installment sale treatment, the same $600K total generates annual tax obligations as payments are received — dramatically improving after-tax cash flow in retirement years.
Payments are funded directly from the business's own post-automation cash flow. After automation, a $1.2M revenue business generates $800K+ in annual cash flow. Even the largest installment payment ($85K/year in Accelerated structure) is less than 11% of automated revenue — comfortably funded from operations.
Why three tiers: A 67-year-old seller wants different cash flow than a 59-year-old. Accelerated serves sellers who need income earlier (healthcare costs, debt payoff). Legacy serves sellers building multi-generational wealth. Standard serves the middle — balanced duration with inflation protection built in.
After all installment sale payments are funded, surplus cash flow across the entire portfolio is distributed as a revenue royalty (not a profit-sharing arrangement, avoiding partnership characterization). Each seller receives a proportional share based on their business's revenue at time of acquisition. This is structured as a contractual royalty on portfolio revenue, not a security interest.
Payments are quarterly (not annual), with a contractual waterfall structure and minimum floor of 10-15% of the installment payment. Many sellers prefer quarterly payouts over annual distributions, as this better aligns with retirement income planning needs. Independent CPA verification ensures transparency. This is the growth upside — what makes the Endowment economically superior to a traditional sale.
The pitch to every seller is identical: "We're not replacing you with a robot. We're replacing you with a dedicated account manager backed by AI that handles all operational work. Your clients get better service — a human who's always available, backed by systems that never miss deadlines. You receive at least what a business broker would get you, paid as predictable annual income, inheritable by your family. And you'll likely receive significantly more."
| Traditional Sale | 10yr Installment | 10yr Growth Dividend | Total | |
|---|---|---|---|---|
| Base case | $699K | $466K | $676K | $1,143K (1.6x) |
| Bull case | $699K | $466K | $888K | $1,354K (1.9x) |
| Bear case | $699K | $466K | $493K | $959K (1.4x) |
Even in the bear case, the Year-1 seller receives 1.4x their traditional sale price over 10 years. The installment sale payments alone cover 67% of the traditional price. The remaining 5 years of installment payments (years 11-15) push every seller well past breakeven regardless of growth dividend performance.
Smart sellers know about ESOPs, MBOs, and seller financing. The competitive set isn't other AI companies — it's other ways to exit a business. Here's how The Owners Endowment stacks up against ALL options on the dimensions sellers actually care about:
| Exit Option | Total Proceeds (10yr NPV) | Seller Risk | Income Duration | Legacy | Tax Treatment |
|---|---|---|---|---|---|
| The Owners Endowment | $1,143K (1.6x FMV) | Low (portfolio-backed) | 15yr, inheritable | Business continues | IRC 453 deferred gains |
| Broker sale to 3rd party | $420-520K (0.7-0.9x after tax) | None (cash upfront) | Lump sum | Business likely dies | Immediate cap gains (37%+) |
| Seller-financed sale | $600K (1.0x FMV) | High (buyer default risk) | 5-7 years | Depends on buyer | IRC 453 installment |
| MBO (Management Buyout) | $600-800K (1.0-1.3x) | High (employees rarely qualify for financing) | 7-10 years if successful | Continuity likely | Varies by structure |
| ESOP | $600-750K (1.0-1.25x) | Medium (ESOP structuring complex, 2+ years) | Lump sum or installment | Employees own business | IRC 1042 rollover if qualified |
Key insights from comparison:
When a 62-year-old business owner's financial advisor asks "what about an ESOP?", the answer isn't "that won't work." The answer is "an ESOP gets you 1.0-1.25x over 2+ years of complexity. The Endowment gets you 1.6x with better tax treatment, less setup time, and a structure your heirs can inherit. Here's the comparison table."
Legal structure note: The IRC Section 453 installment sale structure combined with revenue royalty treatment avoids securities classification under the Howey test. This is not a passive investment expecting profits from others' efforts — it's a sale of a business with seller financing, where the "financing" happens to be structured as installment payments plus a revenue-based royalty. Securities counsel review required before scaling, but initial legal analysis (February 2026) indicates this structure avoids Reg D / Reg A+ filing requirements that would apply to traditional profit-sharing or equity arrangements.
Each acquired business runs on two layers. The agent layer handles 60-70% of operations: order processing, invoicing, vendor coordination, inventory tracking, reporting, compliance, data entry, scheduling, and email triage. Agents are genuinely better here — faster, fewer errors, 24/7, no training ramp. The human layer handles everything client-facing: account management calls, contract renewals, issue resolution, new business development, and quarterly check-ins. Clients won't accept a chatbot managing their vendor relationships, and they shouldn't have to.
In practice: one account manager per 8-10 businesses in the same category, handling only the client relationship layer. Everything behind them is agents. A portfolio of 16 businesses requires 2-3 account managers plus a small ops team — not zero humans, but a fraction of what these businesses required under their original owners.
The technology platform that operates the portfolio is built in three layers: universal modules (client portal, invoicing, email triage — work identically across all categories), category templates (data models and workflow patterns specific to procurement vs. insurance vs. freight), and business configuration (vendor lists, pricing rules, client contacts — data loaded during transition, not code).
The architecture includes autonomous agents with heartbeat-driven execution (similar to openclaw's approach), push-based agents integrated directly into email and other communication channels, and specialized agent skills for core operational jobs. An AI memory layer automatically captures company context as a knowledge vault that compounds over time — every transaction, exception, and decision enriches the operational intelligence available to future automation.
The architecture is designed so marginal engineering cost drops to near-zero as the portfolio scales. Each new business is a configuration task, not an engineering project. Acquisition #1 in a category takes 6 months to automate. Acquisition #15 takes 2 weeks — same universal modules, same category template, just load the business data. And critically, Layer 3 isn't static: every transaction enriches the operational knowledge base, so the system learns how businesses actually operate over time. (See Appendix B for technical architecture details.)
The challenge isn't building agents that work in a demo. It's building agents that work reliably enough that account managers trust them with real client transactions. The Agent OS isn't designed for 100% automation — it's designed for confident partial automation with graceful degradation.
Confidence-based routing: Agents don't just execute tasks — they return a confidence score (0-100) with every decision. "Process this standard PO" = 95% confidence, agent executes automatically. "Handle this unusual vendor substitution request" = 35% confidence, routes to account manager for review. No binary "can the agent do it?" — instead, "how certain is the agent that it's doing it correctly?" Confidence thresholds are tunable per business and adjust over time as agents learn.
Transaction audit trails: Every agent action logs the full decision chain — what data it read, what rules it applied, what alternatives it considered, and why it chose the final action. When something goes wrong (and it will), account managers can see exactly what the agent was thinking. This isn't just debugging infrastructure — it's how the system learns. Exceptions that require human intervention become training data for future automation.
Circuit breakers: If error rates cross 5% in any category within a week, automation for that category auto-pauses and routes everything to humans until the issue is diagnosed. Prevents cascading failures. Better to lose automation velocity than lose client trust.
Daily reconciliation: Batch jobs run every night comparing agent actions against expected outcomes — invoices sent vs. orders received, vendor confirmations vs. client requests, payment timing vs. contract terms. Catches silent failures that agents miss and clients haven't complained about yet. Most bugs in B2B services aren't caught by the client — they're caught when you reconcile your own records.
Model abstraction layer: The Agent OS doesn't hard-code LLM providers. Swap Claude for GPT-4, Gemini, or future models without rewriting agent logic. LLM capabilities are improving exponentially — the architecture must assume models get better, not that they stay static. What requires human review today might be 95% confident in 6 months.
Account manager capacity (realistic): During the first 6 months of a business transition, an account manager handles 3-5 businesses maximum. They're learning client quirks, training agents on exceptions, and building trust. After 6 months, once agents are handling 60-70% of routine work confidently, one AM can cover 8-10 businesses. This is the steady-state model — but you cannot staff for steady-state from Day 1. The financial model accounts for this ramp.
| Metric | Pre-Automation | Post-Automation |
|---|---|---|
| Revenue | $1,200,000 | $1,350,000 (growing) |
| Agent operating costs | $900,000 (75%) | $189,000 (14%) |
| Account manager (1/8th share) | — | ($10,600) |
| Platform / engineering allocation | — | ($28,000) |
| Growth investment (3%) | — | ($40,500) |
| Seller installment payment (annual) | — | ($47,000) |
| Seller bonus pool contribution | — | Variable (~$75K) |
| Owner profit / net cash flow | $300,000 | ~$960,000 |
VCs pattern-match to "AI moats don't exist." If you lead with technology, they discount the entire thesis. The actual moats are built in reverse order of what most AI companies pitch:
Traditional roll-ups are constrained by deal flow. The Endowment inverts this. Sellers come to us because no one else offers tax-deferred, inheritable retirement income from their business. Every happy seller refers others — every retiring boomer knows five more. The pitch gets more credible as the portfolio backing the installment structure grows.
Critically, deal flow doesn't come from direct marketing. It comes from professional referral networks — business brokers, estate attorneys, accountants, and wealth managers who advise retiring owners. These relationships take 3-5 years to build and compound asymmetrically. A competitor in 2030 can build identical Agent OS technology in 6 months. They cannot build 4 years of trust with the professional network that controls deal flow.
Over time, "The Owners Endowment" becomes the recognized destination for retiring business owners — the way Berkshire Hathaway is the acquirer of choice for family-owned manufacturers. This brand is the primary moat.
Every acquisition captures 20-30 years of decision knowledge that has never been recorded as data. The Agent OS captures this operational intelligence for the first time. Every transaction the agents process, every exception the account manager handles, every vendor substitution and pricing decision enriches a shared knowledge base. After a year, the system knows more about how the business operates than the original owner ever documented — because the owner never wrote most of it down.
Decision patterns transfer across businesses in the same category. Vendor performance data, exception handling precedents, pricing benchmarks — intelligence that businesses in traditional fragmented markets never share. A competitor can hire the same engineers and build the same architecture. They cannot replicate 4+ years of accumulated operational knowledge across 50+ businesses.
Dual-use categories — bookkeeping, insurance — aren't just acquisition verticals. They become portfolio-wide infrastructure. Every business acquired in any category gets automated bookkeeping and insurance management as included services, eliminating $2-5K/month in external costs per business on Day 1.
The shared services layer generates deal flow intelligence: you're inside the financials and risk profiles of hundreds of small businesses, identifying acquisition targets before they hit the market. Your bookkeeping clients ARE your acquisition pipeline. No other category compounds across acquisition vertical + horizontal infrastructure + deal flow intelligence simultaneously.
Category dominance means acquisition #15 in B2B procurement takes 2 weeks to integrate vs. 6 months for acquisition #1 — same universal modules, same category template, just load the business configuration. Each new business is a configuration task, not an engineering project.
Account managers serve 8-10 businesses in the same category AND geography. This enables in-person client meetings, shared vendor relationships, and local market knowledge. A concentrated portfolio in "Bay Area B2B procurement" or "Seattle insurance agencies" is operationally more efficient and creates local brand recognition faster than a dispersed national footprint.
Geographic concentration also unlocks growth. Post-acquisition, businesses gain capacity they never had: a dedicated account manager who can expand wallet share with existing clients, restart outbound, and cross-sell across the portfolio. A $1.2M business flat for 5 years under a maxed-out owner can grow 7-10% annually when given dedicated attention. Suppressed growth potential is the hidden asset.
The automation platform makes each new acquisition faster to integrate and cheaper to operate. The three-layer architecture (universal modules + category templates + business configuration) means marginal engineering cost drops to near-zero as the portfolio scales. Confidence-based routing, audit trails, circuit breakers, and daily reconciliation ensure production-grade reliability.
But technology is the weakest moat. A well-funded competitor can build equivalent automation infrastructure in 6-12 months. What they cannot build quickly: the seller brand that generates deal flow, the operational data corpus that makes automation actually work in production, the category depth that enables 2-week integrations, and the geographic concentration that makes account management economics viable.
The competitive reality: An "AI roll-up" that leads with technology will struggle with deal flow, burn capital on custom integration work, and fail to build seller trust. The Owners Endowment leads with brand, uses operational data as the actual competitive advantage, and treats Agent OS as enabling infrastructure. The moats compound in this specific order for a reason.
Ideal businesses are information businesses disguised as services — the owner's value is coordination knowledge, not physical labor. Six filters:
| Filter | What It Means |
|---|---|
| Information layer | Owner coordinates, doesn't do physical work |
| Recurring revenue | Existing contracts, not project-based |
| Structured workflows | 80%+ can be written as "when X, do Y" |
| Thin relationship layer | Client touchpoints are quarterly/annual, not daily — an account manager can step in |
| Suppressed growth potential | Revenue flat because owner is at capacity, not because market is tapped out |
| Large fragmented market | Thousands of similar businesses exist |
Not all categories are equal. Some — like freight brokerage — are pure verticals: acquirable and automatable, but mastering them only helps within that category. Others are dual-use: they work as standalone acquisition categories AND as shared infrastructure that serves every business in the portfolio regardless of category. This distinction drives sequencing.
Every business, no matter what it does, needs bookkeeping, insurance, IT support, and phones answered. If you master one of these categories first — by acquiring and operating real businesses in it — that expertise becomes a service layer deployed across every future acquisition for free. A procurement business you acquire in 2028 doesn't need to keep paying its $3K/month bookkeeper if your portfolio already has automated bookkeeping infrastructure built from running actual bookkeeping practices.
| Category | US Businesses | Automatable | Horizontal Value | Role |
|---|---|---|---|---|
| Bookkeeping / payroll services | ~100,000 | 85%+ | ★★★ | Foundation. Every business needs it. Acquirable vertical + portfolio-wide infrastructure + deal flow intelligence (see below). |
| B2B procurement intermediaries | ~50,000 | 85%+ | — | Proof of concept. Have a free first deal (ExecuSystems) and deep domain knowledge. Pure vertical. |
| Independent insurance agencies | 38,000+ | 80%+ | ★★ | Second dual-use category. Every portfolio company needs coverage. See every client's risk profile and revenue — strong deal flow intelligence. |
| Freight / logistics brokers | ~30,000 | 80%+ | — | Pure vertical. Matching and coordination layer. |
| Property management companies | 300,000+ | 75%+ | — | Pure vertical. Recurring fees, coordination-heavy. |
The bookkeeping trifecta. Bookkeeping is the only category that simultaneously serves as (1) an acquisition vertical in its own right — buy practices, automate them, pay owner guarantees; (2) horizontal infrastructure — every business in the portfolio gets automated bookkeeping as a service, improving unit economics by $2-5K/month per acquisition; and (3) a deal flow pipeline — you're inside the financials of every bookkeeping client, seeing who's profitable, who's slowing down, who's approaching retirement. Your bookkeeping clients ARE your acquisition targets. No other category compounds across all three dimensions.
Acquire dual-use categories first. Bookkeeping and insurance don't just generate their own returns — they reduce the cost and risk of every subsequent acquisition in any category. A procurement business acquired after bookkeeping infrastructure exists is cheaper to operate (no external bookkeeper), better understood (you already see its financials), and better protected (you already manage its insurance). The sequence isn't arbitrary — it's a strategy where each layer makes the next layer more valuable.
Go-to-market: ExecuSystems first, then acquisitions. ExecuSystems validates the ownership transition model and procurement Agent OS in 2026 (free, zero career risk). Bookkeeping acquisition optional in 2026, likely in 2027 — first planned acquisition happens AFTER ExecuSystems validates. This de-risks the core thesis before deploying any capital.
The "AI roll-up" playbook — buy small businesses, automate with AI, expand margins — is popular right now. Most will fail because they're technology-first: "we have AI, what can we buy and automate?" The Owners Endowment is seller-first: "retiring owners need a better option, and AI happens to make one possible." That's a completely different starting point that leads to different decisions about what to buy, how to operate, and how to grow.
The structural differences: roll-ups offer lump-sum buyouts and strip costs. We offer inheritable annual income and invest in growth. Roll-ups are faceless PE vehicles. We're building a brand that retiring owners trust. Roll-ups optimize for exit multiples. We optimize for cash flow durability. The competitive set for deal flow isn't other technology companies — it's other buyers. And the Owners Endowment pitch is a structurally better offer for the seller.
Flexport is the strongest possible case for "a tech company will eat these intermediaries alive." Well-funded ($2.8B raised), 12 years old, best-in-class technology in one of the most digitizable intermediary categories (freight brokerage). After 12 years and $2.8B: roughly 4% market share of a $55B U.S. market. Still not profitably until they sold Convoy assets for $250M. Valuation cut in half from $8B to $3.8B.
And freight brokerage is the easy case for disruption — it's literally a matching marketplace. B2B procurement intermediaries and insurance agencies are more relationship-driven, more fragmented, and more resistant to disruption. There is no Flexport equivalent in these categories because the market is too fragmented and too small for venture economics.
Three structural reasons vertical disruptors don't kill fragmented services businesses:
They target the top of the market. Flexport wants Nike and Amazon shipping thousands of containers internationally. They don't want the regional furniture manufacturer shipping 20 truckloads a month. The unit economics don't work for tech disruptors at the bottom of the market — the same reason these businesses are investable for us is why they're invisible to Flexport.
They fight each other, not the incumbents. Flexport competes with C.H. Robinson, Kuehne+Nagel, DHL. Not with a guy in Tulsa running a 6-person freight brokerage. The local broker's clients aren't comparison shopping on Flexport. They have a 15-year relationship with a person. The switching trigger isn't "better technology" — it's "my broker retired and nobody answered the phone."
They actually make local businesses more defensible. Better technology platforms become infrastructure that our businesses use, not competitors that replace them. The intermediary's value isn't the technology — it's being the person who handles everything so the client doesn't have to think about it.
The real competitive threat is other buyers acquiring the same businesses. This is where the Endowment pitch differentiates: inheritable annual income vs. lump-sum buyout. When a 62-year-old business owner decides who to sell to, our pitch is the one that lets them sleep at night.
The founder builds Agent OS, manages clients, sources deals, and self-funds the first 3-5 acquisitions from personal capital. No investors, no fund structure, no LP governance. 100% ownership of all cash flows. The portfolio self-funds further acquisitions from operating cash by Year 2-3.
| Year | Businesses | Revenue | Team | Annual Cash Flow | Cumulative |
|---|---|---|---|---|---|
| 2026 | 1 | $1,322K | 1 (founder) | $872K | $872K |
| 2027 | 2 | $1,807K | 1 | $856K | $1,727K |
| 2028 | 4 | $3,808K | 2 | $1,669K | $3,397K |
| 2029 | 6 | $5,814K | 2 | $3,053K | $6,449K |
| 2030 | 9 | $8,449K | 3 | $4,310K | $10,759K |
| 2031 | 12 | $10,403K | 3 | $5,732K | $16,491K |
| 2032 | 16 | $13,367K | 3 | $7,551K | $24,043K |
Model assumes front-loaded 20%→10%→4% client attrition, 10% integration failure rate, growth investment, and human staffing costs. Monte Carlo simulation with expert PE feedback.
Understanding where capital comes from — and being honest about overhead — is critical. The model shows acquisition capital deployed over 7 years, but this requires careful accounting of what comes from John's literal bank account vs. what the business generates.
| 2-Year Gate (2026-2027) | Full 7-Year Path (2026-2032) | |
|---|---|---|
| Capital from personal bank transfers | $40-280K | $40-280K |
| 2026 bookkeeping down payment (optional) | $0-40K | $0-40K |
| 2027 procurement acquisition (conservative) | $0-240K* | $0-240K* |
| Capital from business cash flow | $0-240K* | $3.27M-$3.51M |
| 2027 procurement (optimistic scenario) | $0-240K* | $0-240K* |
| 2028-2032 acquisitions (fully self-funded) | — | $3,269K |
| Total capital deployed | $240-280K | $3.55M |
* The 2027 procurement acquisition (~$240K) is funded either from personal savings (conservative) or from business cash flow (optimistic). Personal capital at risk: $40-280K maximum through 2027. From 2028 onward, all acquisition capital comes from portfolio cash flow.
At some point between business #10 and #15, the founder hits the ceiling of what one person can manage. That's when to raise — not because the business needs money (portfolio cash flow funds new acquisitions), but because it needs organizational infrastructure.
The fundraise story is: "I've done this 12+ times with my own money. Here are 4+ years of audited results. The portfolio generates $10M+ in revenue at 60% margins. I want to do it 100 more times." A $10-20M fund deploys into 30-50 additional businesses, hires a deals team, scales the AM layer, and enters a second category.
| Bear | Base | Bull | |
|---|---|---|---|
| Portfolio (Year 10) | 46 businesses | 75 businesses | 100 businesses |
| Revenue | $37M | $76M | $122M |
| Net margin | 43% | 51% | 58% |
| Fund MOIC | 6.2x | 13.8x | 24.9x |
| IRR | 28% | 43% | 57% |
The fund isn't necessary for the business to work. It's an accelerant. And raising from a position of proven results, portfolio cash flow, and 100% ownership gives dramatically better terms than raising from a pitch deck.
ExecuSystems transition (father's B2B procurement business). Build Agent OS v1 on evenings and weekends. Automate Boot Barn workflows ($437K revenue) and Amazon operations ($90K revenue). Working alongside dad, not taking over — building automation, taking modest fees for the value add. Validates: can agents handle procurement operations, and does Boot Barn stay through an AI-powered ownership transition.
Boot Barn is a 25+ year anchor client. If they stay, the core thesis holds. If they leave, the model is dead before any real capital is deployed. Cost: Zero. Free first deal. Zero career risk. Canva salary and equity continue vesting.
2-year decision gate: If not making $1M+/year by end of 2027, doesn't make sense to continue given Canva opportunity cost. This is an explicit decision point, not an indefinite experiment.
If Boot Barn stays and ExecuSystems is operating successfully with AI automation, then deploy capital for first acquisition. Bookkeeping acquisition optional in 2026, likely in 2027. First planned acquisition happens AFTER ExecuSystems validates the core model — don't over-commit to 2026 timeline.
Likely first acquisition: small bookkeeping practice ($25-40K down, Bay Area, 15-20 clients). Bookkeeping is dual-use — acquisition vertical + portfolio infrastructure + deal flow intelligence. Validates recurring revenue model, builds first horizontal infrastructure layer, and starts generating acquisition targets from Day 1 (your bookkeeping clients ARE your pipeline).
Founder builds Agent OS, manages all businesses' clients, handles operations. Total out-of-pocket: ~$25-40K (bookkeeping) or ~$230K (if adding second procurement business). Cash flow begins building.
Decision point: stay at Canva and continue part-time, or go full-time based on data. If not hitting $1M+ annual income by end of 2027, reassess against Canva opportunity cost.
Acquire 2-4 more businesses. Agent OS category template is mature for procurement — new businesses onboard in weeks, not months. Hire first account manager when client calls exceed founder capacity (~business #3-5). Cash flow: $1.7-3.1M annually.
Enter insurance (second dual-use category). Every portfolio company now gets automated bookkeeping AND insurance management as included services. Two horizontal layers compounding across all acquisitions. Insurance client book provides a second deal flow pipeline — seeing risk profiles and revenue across hundreds of businesses.
Begin conversations with 50+ retiring business owners across target categories. Build pipeline for potential fund raise.
Portfolio generating $8-10M revenue. Second account manager, possibly a junior engineer. Cash flow: $4-6M annually. Portfolio self-funds all new acquisitions from operating cash flow.
Decision: continue self-funded at ~3-4 acquisitions/year, or raise a fund to accelerate into second category and scale to 50+ businesses.
Self-funded path: 16+ businesses, $13.4M revenue, $7.6M annual cash flow, 3-person team. Continue acquiring 3-4/year indefinitely.
Fund path: 50-75+ businesses, $50-80M+ revenue, multi-category. Agent OS deeply defensible. Inbound deal flow exceeds capacity. Paths: private cash flow machine, permanent capital vehicle, or IPO the holding company.
Front-loaded attrition is real: 15-25% of revenue at risk in Year 1, dropping to 3-5% by Year 3. For single-client businesses, it's binary. Mitigation: Dedicated account manager overlaps with original owner 3-6 months — clients experience an upgrade, not abandonment. Only acquire businesses where the relationship layer is thin and standardized. The pitch to clients: "Nothing changes. You still have a person to call. Behind the scenes, everything is faster and more reliable." Validated with ExecuSystems before scaling.
Error tolerance in B2B procurement is near zero. A wrong order can cost a client relationship. Mitigation: Human-in-the-loop for all high-value transactions in Year 1-2. Progressive automation as reliability improves. Never fully remove human oversight for exception handling.
~10-15% of businesses will have workflows that resist automation — legacy systems, undocumented processes, edge cases requiring constant human judgment. These run at 2-3x expected opex indefinitely. Mitigation: Modeled at 10% failure rate. Failed integrations still generate revenue and cover their installment payments — they just don't contribute the same margin. Due diligence evaluates workflow complexity before acquisition.
By acquisition #15, carrying 15 simultaneous installment payment obligations. Mitigation: Each installment payment funded from its own business's cash flow — the $47K/yr payment is less than 5% of automated revenue. Model maintains 12 months of payment reserves plus $500K minimum cash. Need actuarial expertise before scaling past 20 businesses.
If AI gets good enough, clients could handle procurement/insurance/freight directly without intermediaries. Mitigation: The clients of these businesses — regional retailers, mid-market manufacturers, small commercial buyers — are the slowest technology adopters in the economy. Enterprise AI adoption is measured in decades, not quarters. The self-funded path extracts $10-24M before this risk materializes. And the client portal in Agent OS is the path to evolving the product if the landscape shifts.
Self-funded model makes the founder the single point of failure: engineer, operator, deal sourcer, client manager simultaneously. Mitigation: This is explicitly the plan for businesses #1-3, with structured hiring triggers. Part-time AM at business #3, full-time at business #5, junior engineer at business #8. The Agent OS architecture is designed so that the founder's engineering work compounds — each module built once serves every future business.
Three trends converge for the first time: the boomer retirement wave is a demographic certainty already underway. AI agent capabilities have crossed the threshold where multi-hour autonomous business workflows are reliable. And the gap between these capabilities and practical deployment is closing — 2026-2028 is when agents become production-grade for structured B2B operations.
The intersection of "can personally build the entire Agent OS," "has built and sold a company," and "has a free first deal" is an extremely small set. The founder led Canva's Magic Studio (TIME Best Invention 2024), managing 230+ engineers building production AI systems for 200M+ users. Two-time Y Combinator founder. Previous company (URX) acquired by Pinterest. Has operated at Google, Pinterest, and Canva.
Critically: has a family business (ExecuSystems) available as a zero-cost proof of concept, with an existing anchor client relationship to validate the hardest assumption in the model. And has the personal capital (from Canva equity) to self-fund the first 3-5 acquisitions without external investment.
In a world where AI capabilities are accelerating exponentially, the self-funded path is the strategically correct structure. It extracts cash immediately (no fund waterfall diluting returns). It provides total flexibility (no LP governance, no fund timeline). It hedges against uncertainty (if the landscape shifts, the founder has already extracted $10M+ and can pivot). And it creates the strongest possible fundraise position if and when scale requires outside capital.
The explicit threshold: by the end of Year 2 (late 2027), is the portfolio generating at least $1M/year in annual income? This is the break-even point where continuing full-time makes more sense than returning to Canva.
| Outcome | Probability | Year 2 Income | Decision |
|---|---|---|---|
| Doesn't work | 35% | $400-700K | Return to Canva. Keep ExecuSystems as $400-700K/yr side business. Total opportunity cost: ~$300K (2 years of foregone Canva equity appreciation). |
| Works marginally | 15% | $850K-1.2M | Borderline. Either continue cautiously (no new acquisitions until income stabilizes above $1M) or return to Canva with larger side business. |
| Works clearly | 50% | $1.2M-1.8M | Continue. You're making more than Canva base + equity, with better trajectory. |
The massive career opportunity cost: Staying at Canva through 2032 means $9.625M in total comp (7 years × $1.375M annual salary + equity). Leaving in Year 2 to pursue The Owners Endowment full-time means giving up $7.9M of that total — potentially the highest-value equity you'll ever vest. This is not a casual decision. The portfolio must be generating $1M+/year by Year 2 to justify the trade, and the trajectory must credibly point to $3M+/year by Year 5 to make the opportunity cost worth it.
These show the complete probability distribution across all scenarios, including those that fail at Year 2. The probabilities below represent absolute likelihoods, not conditional outcomes.
| Outcome | Probability | Description | Year 2 Income | Year 7 Income | 7yr Cumulative |
|---|---|---|---|---|---|
| Doesn't work | 35% | Client attrition kills model. Return to Canva after Year 2. ExecuSystems continues as side business. | $400-700K | n/a | $2-5M |
| Works marginally | 15% | Passes Year 2 gate but stays small. Grows to 6-10 businesses. Solid personal income. | $850K-1.2M | $3-5M | $8-15M |
| Works clearly | 35% | Strong Year 2 performance. Grows to 12-16 businesses. Profitable personal cash machine. | $1.2-1.8M | $5-7.6M | $15-24M |
| Raises fund, scales | 12% | Exceptional Year 2 results. Self-funded proof → fund raise → 50-75 businesses. | $1.5-2M | $8M+ | $30M+ (cash + equity) |
| Generational company | 3% | Everything works perfectly. Multi-category, 100+ businesses. Permanent capital vehicle or IPO. | $2M+ | $15M+ | $100M+ (equity) |
The 35% "doesn't work" scenario is NOT catastrophic failure. It's: ExecuSystems transitions successfully (Boot Barn stays, generates $400-500K/year in cash flow), but the second and third acquisitions experience 40%+ client attrition in Year 1. By late 2027, the portfolio is generating $600-800K/year — good, but not $1M+. The rational decision is to return to Canva, likely with a promotion (you've been gone 18 months, Canva wants you back), keep ExecuSystems running as a high-margin side business, and extract $400-700K/year indefinitely with near-zero time commitment.
Total opportunity cost in the "doesn't work" scenario: ~$300K (18-24 months of foregone Canva equity appreciation). You leave with a profitable side business generating more than most people's salary. This is the floor, not the ceiling.
The critical insight: the downside is capped and acceptable. The upside is uncapped and massive. If it doesn't work by Year 2, you return to one of the best jobs in tech with a $500K/year side business. If it works, you're on a path to $10-100M+ over 7 years with 100% ownership. The 2-year validation period costs you at most $300K in opportunity cost. The 50% probability that it works generates $10M+ in the base case.
You don't need the generational outcome to justify the decision. You need the Year 2 income to hit $1M+. Everything after that is compounding on an already-validated model.
The entire thesis rests on one critical question: does the anchor client stay through an ownership transition powered by AI agents and a dedicated account manager? ExecuSystems tests this in 2026 — zero cost, zero career risk, while still employed at Canva. If Boot Barn stays, deploy capital for first acquisition in 2027. If not, the thesis is dead before any real capital is at risk.
The critical asymmetry: in the base case (30% probability), the self-funded path generates $10-24M over 7 years. The downside case (35% probability) still extracts $2-5M. You don't need the platform outcome to build generational wealth. The platform outcome is leverage on an already-excellent base case.
Traditional acquirer pitch: "I'll pay you 2x EBITDA and figure out how to run your business." Our pitch: "I'll pay you from your business's own cash flow, guaranteed and inheritable, because I've built the system that lets it run without you."
That's not just a better offer. It's a different category.
The technology platform that operates the portfolio is built in three layers, designed to scale across categories without bespoke engineering per business:
Client portal (order status, requests, invoices), invoicing engine, email/comms triage and routing, exception escalation to account managers, reporting dashboards, document management, scheduling. These work identically across procurement, insurance, freight, property management. Every business gets the full stack.
Data models, workflow patterns, and domain-specific logic. "B2B procurement" defines POs, vendors, catalogs, margins, fulfillment tracking. "Insurance agency" defines policies, carriers, renewals, claims. One template covers every business in a category with light configuration.
Vendor lists, client contacts, pricing rules, margin tables, approval thresholds. Loaded during transition from whatever the owner was using — spreadsheets, email folders, their memory. This is the 2-6 months of integration work per business. If you're writing code for a specific business, something is wrong. Critically, Layer 3 isn't a one-time data load. Every transaction the agents process, every exception the account manager handles, every vendor substitution and pricing decision enriches the operational knowledge base. After a year, the system knows more about how the business actually operates than the original owner ever documented — because the owner never wrote most of it down.
The client-facing portal is a core module, not an afterthought. It's the same reason you have a Shopify dashboard — different data for different stores, same underlying system. It increases switching costs (clients integrate it into their workflow), improves the client experience (self-serve status checks instead of phone tag), and reduces the account manager's workload to pure relationship management.
The heuristic: Agent OS builds anything that follows the pattern "every business in every category needs this." It does NOT build anything where the logic is "this specific client needs this specific thing." Capabilities, not customizations. If a business requires custom code, it either means a module is missing (fix the platform) or the business doesn't pass the acquisition filter (don't buy it).