RaaS (Results as a Service) means your client pays when your work produces a result, not for your hours or for a tool.
Inside: every headline behind the shift, explained in plain words with the source. How RaaS works and how to price it. And 12 offers for different kinds of agencies, each with what counts as a result, a worked price and a pitch line.
Part 1
In February 2026, investors started dumping software stocks. By May, the two biggest AI labs had both launched services companies. Here's each headline and what it means for you.
Anthropic, the company behind Claude, released an AI tool for in-house lawyers. Shares of legal software and publishing firms tumbled the same day. A trader at the investment bank Jefferies told Bloomberg: "We call it the 'SaaSpocalypse,' an apocalypse for software-as-a-service stocks."
Axios reported the same slide: Anthropic's release triggered a sell-off across software stocks.
In two days, hundreds of billions of dollars were wiped off the stocks, bonds and loans of tech companies. Software was at the center: the software stocks in one iShares fund lost almost $1 trillion in value over seven days.
Bloomberg: Trillion-Dollar Tech Wipeout Ensnares All Stocks in AI's Path
Anthropic's new tools for its Claude "Cowork" agent handle work that software and data companies sell as core products: legal research, CRM, analytics. The S&P 500 Software & Services Index fell over 4% in a day, its 8th losing session in a row, and was down about 20% for the year. Hedge funds had already bet about $24 billion against software stocks. Thomson Reuters, Salesforce and LegalZoom were among the hardest hit.
By the end of February: Salesforce down 21% for the year, ServiceNow 26%, Adobe 22%, Intuit 37%. The fear: companies buy fewer software seats when AI does the tasks the seats were for. Mistral's CEO told CNBC that more than half of the software in a company could be replaced by AI. An investment director at GAM said the software model "in its current form is impaired."
CNBC: Software was going to eat the world. Now it's facing an 'existential' crisis
Julien Bek, an investor at Sequoia (one of the most famous venture firms), wrote that the next trillion-dollar company will be "a software company masquerading as a services firm." In plain words: AI does the work, and the customer buys the finished work, not a tool. The post passed 1 million views on X within days, and Fortune covered it in April.
Fortune: This Sequoia partner thinks AI-enabled services are the new software
HubSpot switched two AI agents to paying per result. Its Customer Agent went from $1 per conversation to $0.50 per resolved conversation. Its Prospecting Agent went from a monthly charge per contact to $1 per lead recommended for outreach, meaning you pay when a prospect gets qualified and handed to your team. HubSpot says the Customer Agent resolves 65% of conversations for 8,000+ customers. Its Chief Customer Officer: "You pay when it works, full stop."
Intercom prices its Fin AI agent from $0.99 per outcome: a customer question Fin fully resolves. No resolution, no charge for that ticket (Intercom still has seat plans on top).
Anthropic, Blackstone, Hellman & Friedman and Goldman Sachs formed a new company that works with mid-sized businesses (think community banks, manufacturers, regional health systems) to put Claude into their most important work. Engineers sit with the customer, find where AI helps most, build the system and support it long term. More investment firms back it, including General Atlantic, GIC and Sequoia Capital.
OpenAI set up the OpenAI Deployment Company with more than $4 billion from 19 investment firms led by TPG (with Advent, Bain Capital and Brookfield). It puts OpenAI engineers inside client companies to get AI from pilot into daily work. To staff it fast, OpenAI is buying Tomoro, an AI consulting firm whose clients include Mattel, Red Bull, Tesco and Virgin Atlantic. OpenAI's COO: "Our customers tell us they need help going from pilot to production."
AI made software cheap, so tools are losing their pricing power. Results are still rare, so the money is moving to whoever delivers them. The AI labs are building services companies to catch it. You already run one.
Part 2
You sell a result the client can count. AI does most of the work, so your cost per result drops, and you keep the difference.
| Retainer | Software (SaaS) | RaaS | |
|---|---|---|---|
| Client pays for | Your hours | Access to a tool | A result |
| Who does the work | Your team | The client | You, with AI |
| Client's risk | High: pays even if nothing works | High: pays even if no one uses it | Low: no result, no bill |
| Your upside | Capped by hours | n/a | Grows as AI cuts your cost per result |
Every price on this page is a worked example, not a market rate. Set yours from the client's numbers and your own costs.
Part 3
Find your kind of agency, then open the offer for the details. Pick one. Don't build them all.
| If you run a… | Sell… |
|---|---|
| Lead gen / cold email agency | Held sales calls |
| Paid ads / performance marketing | Qualified leads |
| SEO agency | Page-1 rankings that hold |
| Email / e-commerce marketing | Recovered revenue |
| Social media agency | Qualified inbound DMs |
| Design agency | Approved designs in 48 hours |
| Web / CRO agency | Conversion lift |
| Video / content studio | Ready-to-post videos |
| AI automation / support | Resolved tickets |
| Recruiting agency | Hires who stay 90 days |
| Bookkeeping / finance ops | Books closed on time |
| PR agency | Earned media placements |
Qualified sales calls that actually happen on your client's calendar.
Leads from your campaigns that pass the client's own qualification rules.
Agreed buying-intent keywords on Google's first page, kept there.
Sales from abandoned carts, win-back and post-purchase flows you build and run.
Real potential buyers who message the client because of the content you make.
Finished, client-approved designs (ads, social posts, landing sections), billed only when approved.
More sales or sign-ups from the same traffic, proven with an A/B test.
Short videos edited, captioned and approved, ready to post, paid per approved video.
Customer support tickets solved without a person stepping in. (Intercom and HubSpot already price this way.)
Placed candidates, paid in full only if they're still there after 90 days.
The client's books, reconciled and closed every month by a set date.
Articles, interviews and podcast spots in agreed outlets, paid per placement that runs.
Part 4
Start with clients who already trust you. A 30-day pilot gets a yes faster than a cold pitch.
Sources
Prices in the offers are worked examples, not market rates. Quotes and figures are as reported by each source on the date shown.