Anthropic’s PE-Backed Venture, Ode, Pushes to Get Paid for AI Results

AI Summary

Ode, a venture backed by private equity firms and Anthropic, builds custom AI tools for businesses and typically charges fixed project fees. It is adding performance-based fees tied to clients’ financial results, while facing challenges in agreeing on how to measure AI-driven gains; its acquired predecessor Fractional AI developed a tool that shortened LogicGate’s onboarding process.

Private equity firms and AI labs are sorting out how to charge clients and hire enough engineers for joint ventures they’ve launched to build custom AI tools for businesses. That comes as private equity firms are looking to the technology to boost profits, including at companies they’ve struggled to sell. One such team-up is Ode with Anthropic, which launched in July with $1.5 billion in capital commitments from investors including Blackstone and Hellman & Friedman. Ode currently charges most of the several dozen businesses it works with a fixed fee to develop and launch custom AI products, people familiar with the partnership say. The company lets clients retain ownership of the AI software it builds, the people said, allowing them to keep using or even sell the tools once completed without paying Ode ongoing fees.  Those terms help Ode attract customers that balked at the terms offered by the team of engineers at Anthropic that embeds with enterprises, which seeks to retain ownership of the custom software it builds and license it back to customers, one of the people said. But Ode has been looking at other ways to earn more on AI projects.  Some Ode executives initially proposed coinvesting in companies that were also its customers, but Blackstone’s team pushed back on the idea, one of the people said. Instead, Ode decided to charge performance-based bonus fees on top of the fixed project fees, allowing it to earn more if clients hit specific financial targets.  Less than 10% of Ode’s clients are paying those performance-based fees, one of the people said, though Ode is aiming to grow that share. Setting the fees can be difficult because the company and its customers both need to agree on how to measure any AI-driven gains, including setting a baseline and financial or product-related goals. The fees may be negotiated once a contract is signed and Ode then does some early exploratory work and collects data, giving both sides enough information to set terms before moving on to the building phase, the people said.  Ode inherited some of its business from Fractional AI, a startup founded in 2024 that the venture acquired in May for an undisclosed sum. Fractional already worked mostly with private equity–backed companies to automate tasks and build custom AI software, and Fractional co-founders Chris Taylor and Eddie Siegel became Ode’s CEO and chief technology officer, respectively.  One of those clients, risk and compliance software company LogicGate, paid a fixed fee in the low seven figures for a project that began under Fractional last summer and finished this fall, said LogicGate CEO Diego Panama. (LogicGate’s backers include growth equity firm PSG, which led the company’s $113 million funding round in 2021.)  Ode engineers built a custom agent that shortened LogicGate’s client onboarding from 120 days to 90 days or less. LogicGate received full intellectual property ownership and training to run the tool independently, Panama said, and it subsequently began selling the agent to other businesses and has signed several deals. The agent is set to launch publicly at the end of the month.  Around half of Ode’s clients are software or business services companies, while the rest come from industrial and other sectors. Its client base is split roughly equally between portfolio companies of Ode’s backers, including at least six Blackstone companies, and other businesses.  All of the Blackstone portfolio companies working with Ode currently pay fixed fees, including accounting firm Citrin Cooperman, where Ode engineers built a new onboarding system, similar to the work they did at LogicGate. Blackstone typically sets targets to improve earnings before interest, taxes, depreciation and amortization in a specific division when evaluating how an AI project is working.  Other AI services firms are experimenting with similar fee structures. Hyde, backed by Bain Capital Ventures and General Catalyst, launched last year and helps businesses build custom AI models using their own data. Clients then own the customer models and can run them themselves or pay Hyde to host them.  Hyde typically assigns two engineers to work with clients for around four weeks per project, charging for that work plus a recurring fee for using its software platform. (Clients also pay for compute based on usage.) Around half its contracted revenue includes a portion of fees tied to results—its fee for an automotive client, for example, is tied to whether agents meet certain car sales goals. Hyde CEO Anirudh Sanga said demand for Hyde’s services has roughly doubled monthly for the past three months.  At the same time, a shortage of engineers is limiting how much business AI services firms can take on. Anthropic itself has sought to address the shortage, announcing a $100 million effort last month to train 10,000 deployment engineers at customer and partner companies by the end of next year.  Upstart AI services firms are also competing with maj