SaaS sells access. RaaS sells the result.
Agents won’t take your sales meeting. They’ll check the numbers and pick whoever delivers. New article on why the next trillion-dollar software company won’t have a sales team ↓ https://x.com/i/article/2108541750759538688
The Next Trillion-Dollar Software Company Won’t Have a Sales Team
This week I was in Singapore for Token2049, and a conversation there hasn’t left me.
I was talking to the co-founder of a very large payments technology company. Somewhere in the conversation he mentioned that getting time with the founder at one of his biggest customers isn’t always easy, because he’s seen as a service provider.
It was a small moment in a long conversation, but it stayed with me. By the time I got back to my hotel, I was already working through where this leads.
Why does this happen even to the biggest, best-run service companies? Because they sell a service, not a result. Access is something a buyer can’t see working, and anything a buyer can’t see is a cost. Cost centers don’t get their calls returned.
So I put a question to people while I was there: what if, instead of selling a service, you sold the result? I asked people at Google, at PayPal, and at a couple of other prominent companies. The reaction was nearly the same every time: yes, that makes sense. Conversations aren’t data, but the data points the same way.
Founders call back the vendors that made or saved them money
The market is already pricing this in. The Dow Jones U.S. Software Index fell about 36% from its Oct. 28, 2025 record to April 10, 2026, on fears of what AI would do to SaaS, then recovered about 42.5% from that low by Aug. 31. Forrester found per-seat pricing adoption dropped from 21% to 15% in twelve months.
Result as a Service
What replaces the seat is the result: you pay for the outcome, not for access. I call it RaaS.
It isn’t theoretical. Intercom’s Fin charges $0.99 per outcome, billed once per conversation, and a reopened conversation refunds the outcome. Fin grew from $1M to $100M+ ARR on this model, backed by a performance guarantee of up to $1M. Salesforce then agreed to acquire it for about $3.6B in cash. Sierra, which also charges per resolved interaction, is valued at $15.8B and went from $100M to $200M ARR in two quarters. Intercom Fin AI Agent 2026: $0.99 per Outcome, Features, Limits
Intercom’s president described the effect well: outcome pricing exposed every weak link, because sales couldn’t optimize for licenses and the product had to work consistently. When you sell results, the result is the meeting. saastr
What an agent looks like when nobody is paid for the result.
I’m one of @MakeMyTrip’s heavy customers. Last year I booked around 32 lakh rupees (roughly $35,000) of travel with them. My booking came with a free date change. When I went to use it, their support agent kept telling me there was no flight on my new date. I checked in the same portal, and the flight was there. When I tried to reach a human, I couldn’t find a way. I won’t book with them again. (MakeMyTrip: this is fixable.)
I don’t know how they measure their agent. I know what it did for me: it ended the conversation without solving the problem, and it cost them a customer worth a lot. I’m not alone. Gartner’s 2026 survey found 87% of customers say a route to a human is essential when AI handles service, and only 27% would try a chatbot again after a bad experience. getmacha
The question for any agent is: is it paid for ending conversations, or for solving problems?
What this looks like for a founder with no finance team
Take a small business founder who is also the CFO. A UK survey of 1,000 SMEs found a third couldn’t correctly define cash flow, and 82% reported cash flow difficulties. Now sell them a result instead of a tool. startups
Idle cash. A service connects to the accounts, learns the minimum liquidity needed, and puts the rest to work. Treasury platforms already sweep excess cash into higher-yielding options while holding the buffers you set. The RaaS version has no subscription: at month end it compares your cash performance with last month at the same sales and takes a share of the improvement. Nothing found, nothing owed. brex
FX. An importer with one bank account has no bargaining power on every payment. Banks typically add a 2–4% markup on cross-border SME payments, which is $10,000–$20,000 a month on $500,000 of flows. Fintech providers typically charge 0.3–0.8%. A layer that routes each payment to the cheapest option gives that money back, and keeps a share as its fee. xtransferxtransfer
That founder doesn’t want software. He wants the bill to go down, and he’ll take the call of whoever makes it happen. And both results are checkable: same sales, this month versus last.
The buyer that never takes a meeting.
Now add the shift: the buyer is becoming an agent. A human CEO relies on what the team reports. An agent pulls the raw data, tests the vendor against alternatives, and acts. It has no friendship with the founder and no appetite for a steak dinner. Agents don’t respond to relationships. They respond to proof.
And you don’t need a giant organization to win. Measurement, integration and building are all getting cheaper, so a small team can build the best result engine in a narrow category and be tested against incumbents from day one.
The co-founder I met in Singapore was pointing at something real. Sell a service, and even a very good one can end up treated like a vendor. Sell results, and you’re a partner, or you’re simply the default because an agent checked the numbers and picked you.
The sales team that wins on a good lunch is about to meet a buyer that doesn’t eat.