Direct answer
OffDeal is the mergers-and-acquisitions case of outcome-vertical integration. It is a Toronto boutique that runs sell-side, and some buy-side, processes for founder-led companies in the $1 to $10 million EBITDA band, roughly $5 to $100 million of revenue, and keeps the mandate. It does not sell a document generator to Houlihan Lokey.
The hiring copy, still live in September 2026, is the thesis in one line: “Instead of selling software to Goldman Sachs, we’re building what Goldman itself would look like if it launched in 2025.” That is the same abandoned-software decision Harper and Crosby made, pointed at a fee that only appears when a deal closes.
Funding and scale, as reported: Y Combinator winter 2024; about $4.7 million seed and a $12 million Series A led by Radical Ventures in July 2025; about $17 million total at a mark around $100 million. The Financial Times, at that raise, had them at roughly ten deals with fewer than ten people, three of them bankers. Later 2026 write-ups put completed or launched sell-side work above 30. Process time claimed around four months versus a six-to-twelve-month boutique slog. Ambition stated to the Financial Times: $100 million of run-rate revenue on 100 deals by the end of 2027. Treat the 2027 line as a target.
On a non-factory rubric set beside Crosby, OffDeal scored 32 out of 40, behind Crosby at 38, Harper at 34, and AllDigital at 34. The gap is accountability and evidence, not the make-or-buy call. A success fee is not malpractice insurance. Thirty deals is not thirteen thousand contracts. The call itself is clean. Crosby is a comparison, not a separate essay on this site.
xvertical does not work for OffDeal. OffDeal is not a client. The company is here only as a case a reader can check.
Where this sits next to the other cases
| Case | Owned outcome | Rented layer |
|---|---|---|
| SpaceX | The vehicle, built in-house | Some outside parts |
| BYD | Cell to car, plus driving silicon | Some foundry |
| Anduril | Lattice, plus enough hardware | Some autonomy software, and the government budget |
| Harper | Bound policy and the customer | Carrier capital |
| AllDigital | Bind or decline | Carrier paper |
| OffDeal | Closed sale and the success fee | Buyer capital, and the seller’s trust |
Owning a plant and owning a professional result are both vertical integration. OffDeal’s result is a signed purchase agreement for a paving company, an HVAC shop, a waste hauler, or a pickleball operator, the Financial Times’ early client list. That is Main Street, like Harper, with a much larger cheque per file.
What vertical means in a two-person deal pod
Traditional lower-middle-market mergers work is a pyramid: analysts build the book, associates run the process, a managing director owns the founder. The pyramid is why a $4 million fee on a $40 million sale can still lose money. Too many hours sit below the partner.
OffDeal’s published design:
- Intake and positioning: the founder, the story
- Book assembly: the sale document, the teaser, the comparisons, the model (Archie, the internal analyst)
- Buyer universe from a proprietary index, not a shared blast
- Outreach and nondisclosure traffic
- Auction and negotiation
- Close, then the success fee
A deal-tech vendor owns a slice of steps 2 to 4 and invoices the boutique. OffDeal owns steps 1 to 6. Coverage in 2026 describes two-person pods, one banker on origination and one on execution, with the system doing the analyst stack. Job posts claim about 80% of analyst workload automated. Eldarov’s older argument, to the Financial Times: the analyst-to-managing-director totem pole creates people whose job is to justify the totem pole. Delete the middle or the unit economics never reach a paving company.
They started, in the Y Combinator launch, as a buy-side sourcing agent. The firm that exists in 2026 is a sell-side bank that still does buy-side work. That pivot is the integration tightening: the scarce relationship is the seller, not the search-fund associate who will not pay a subscription.
Pricing is success-based, often described as no upfront retainer. That aligns the firm with close, the way a per-document fee aligns with finishing a redline. It also concentrates risk. A quarter with no closings is a software company with no seats sold, except the bankers still fly to Philadelphia.
The numbers that matter
| Claim | Figure | Source | Date |
|---|---|---|---|
| Founded and headquarters | 2023–24, Toronto; U.S. mandates | Y Combinator / Radical | — |
| Founders | Ori Eldarov (chief executive, formerly RBC, Harvard Business School); Alston Lin (chief technology officer, Waterloo) | Company | — |
| Y Combinator | Winter 2024 | Y Combinator | 2024 |
| Seed | About $4.7 million | 2026 roundups | 2024 |
| Series A | $12 million, Radical Ventures lead, about $100 million mark | Financial Times | July 2025 |
| Total raised | About $17 to $17.2 million | CB Insights / job posts | 2026 |
| Target company | $1 to $10 million EBITDA; $5 to $100 million revenue | Financial Times / jobs | 2025–26 |
| Deals at the Series A | About 10 closed or near close; fewer than 10 staff, 3 bankers | Financial Times | July 2025 |
| Deals later | 30-plus sell-side launched or completed | 2026 industry notes | 2026 |
| Cycle time | About 4 months versus 6 to 12 typical | Firm positioning | 2026 |
| 2027 target | $100 million run-rate on 100 deals | Chief executive to the Financial Times | 2025 |
| Fees | Success-based; “millions” generated, in the firm’s jobs copy | Company jobs | September 2026 |
| Internal tool | Archie, the analyst system | Engineering write-ups | 2026 |
| Analyst automation claim | About 80% of analyst workload | Jobs copy | 2026 |
| Early sectors | Paving, HVAC, waste, pickleball | Financial Times | 2025 |
How to read the table. The capital table and the Financial Times snapshot are the hard facts. “30-plus deals” is a 2026 secondary figure, not a single audited list of closed sales. “$100 million by 2027” is an ambition stated in 2025. “Millions in fees” is recruiting language. Four-month cycle time is positioning against a stylized traditional process. Do not build a revenue model off the jobs page.
Why this is integration, not a chatbot for sale documents
They kept the mandate. The learning that matters in small-business sales is which buyers actually write cheques in asphalt, and which founders will sit through diligence. That data lives in closed deals. A software tool sees anonymized usage. OffDeal sees the wire.
The pod is the factory. Two people plus Archie is the industrial claim. If the pod cannot run a full process, the system is a demo.
They picked the inventory banks refuse. Eighty percent of sub-$10 million EBITDA sales in private-equity deals happen outside a formal auction, Eldarov wrote in 2025. That is the hard inventory of this story: relationship-heavy, and fee-starved. The system only has a business if that inventory can carry an institutional process at boutique cost.
Engineers outrank the pyramid. Eldarov has said the firm has more engineers than bankers. A buyer graph built by a named engineer is the opposite of a shared broker email blast. Whether that graph is a lasting advantage is unproven. That they staffed it as a product, not as a summer-analyst chore, is the integration.
What OffDeal still does not own
- The buyer’s money. A strong sale document does not close a deal. Credit, sponsor appetite, and founder nerves do.
- A license like Crosby’s. Advisory liability exists. It is not the same as signing a contract as counsel of record, or declining a directors-and-officers risk as a managing general agent.
- Proof at 100 deals. Ten at the Series A, thirty in later chatter, one hundred as a 2027 slide. Do not skip those commas.
- A public evaluation. Crosby shipped Redline Bench. OffDeal has not published “did the buyer list contain the eventual winner.” Archie is internal.
- The buy-side identity. The Y Combinator launch was “an assistant for buyers.” 2026 is “Goldman for small businesses” on the sell side. Both can be true. Only one is the profit and loss of owning the mandate.
- Canada as the market. Radical is in Toronto. Mandates are described as U.S. The firm is a cross-border product company sitting on a local-trust industry.
Same four-part test
| Test | Factory version | OffDeal version |
|---|---|---|
| Expensive layer is scarce | Cell, airframe, underwriter | Banker hours, the buyer graph, and founder trust |
| Volume amortizes the stack | Units off a line | Dozens, then hundreds, of similar small-business processes |
| Model co-designed with the work | Chip next to the motor | Archie graded by the banker who takes the offer call |
| You are the first customer | Own fleet | OffDeal bankers are Archie’s first users |
It fails when a startup sells “AI for your boutique” and the boutique still owns the founder. The traces walk out with the customer.
What operators should copy
- If the buyer already pays on close, be the advisor. Do not be the plugin that drafts the sale document.
- Staff engineers against the hour-eating step: the buyer graph and the book. Staff bankers against the room.
- Price the outcome. Retainers recreate the totem pole.
- Stay inside a band where the incumbents’ cost structure does not work. Here, that band is $1 to $10 million of EBITDA.
- Count closed sales, not prompts.
Do not copy the $100 million 2027 slide. Copy the sentence about not selling to Goldman.
Key takeaways
- OffDeal is outcome-vertical integration in lower-middle-market mergers: own the mandate, own the pod, own the analyst system, rent the buyer’s capital.
- About $17 million raised, and a Series A mark around $100 million in July 2025, from Radical Ventures and Y Combinator. The Financial Times had about 10 deals and fewer than 10 people at the raise. Later figures say 30-plus.
- The fee is on success. The pods are two people. Archie stays in-house. The 2027 target of 100 deals is a target.
- The score is 32 out of 40 on the rubric set beside Crosby: the right make-or-buy call, with thinner liability and thinner evidence than Crosby.
- Harper and AllDigital are on this site. Crosby is not.
Sources
- Financial Times, “Behold the first AI-native investment bank,” 29 July 2025.
- Radical Ventures, OffDeal portfolio note.
- Y Combinator, OffDeal company page (launch positioning).
- CB Insights, OffDeal profile (capital stack).
- OffDeal job listings, September 2026 (pod design, “not selling to Goldman,” fee language).
- Astris Nexus, “AI-Native M&A Firms Are Here,” March 2026 (30-plus processes, two-person pods).
- Ori Eldarov, Forbes Council, January 2025.
The $100 million of 2027 revenue, the 80% analyst-automation claim, and “millions in fees” are firm or founder statements. Deal counts after the Financial Times article are secondary and not a single audited list.
Questions this essay answers
What is OffDeal?
A Toronto AI-native investment bank that advises founder-led businesses, typically with $1 to $10 million of EBITDA, on selling and sometimes buying companies. It charges on success.
Is OffDeal software for bankers?
No. The company says it is building the bank, not a tool for Goldman. Archie, the analyst system, is internal.
Why does that count as vertical integration?
Because OffDeal owns origination, book-building, buyer coverage, the process, and the close fee in one firm. A vendor of sale documents owns only the draft.
How much has OffDeal raised?
About $17 million, a seed near $4.7 million and a $12 million Series A led by Radical Ventures in July 2025, reported around a $100 million valuation.
How many deals has OffDeal done?
The Financial Times had about ten at the Series A. Later 2026 sources say more than thirty sell-side processes launched or completed. There is no public tombstone book in the sources used here.
How is OffDeal different from Crosby?
Same pattern. Crosby takes legal liability on a document. OffDeal takes a success fee on a sale. One file at Crosby is a nondisclosure agreement. One file at OffDeal is a company.
How is OffDeal different from Harper?
Harper binds many small policies. OffDeal closes fewer, larger transactions. Both kept the customer instead of selling the operating system.