Composo: Pioneering Reliable AI Evaluation for Enterprises

Artificial Intelligence (AI) and the large language models (LLMs) that drive its capabilities hold enormous promise for various applications. However, a significant concern remains: reliability. As these models shake up industries, inconsistency raises flags, posing challenges for businesses that rely on them. With no clear resolution in sight, startups are seizing the opportunity to help enterprises ensure their AI applications perform as expected.

One such startup is Composo, based in London. They’re on a mission to enhance the reliability of LLM-powered applications through custom models designed for enterprises. This London-based newcomer stands shoulder to shoulder with companies like Agenta, Freeplay, and LangSmith, all vying to offer robust alternatives to traditional human testing and existing evaluation tools. But Composo sets itself apart with a dual approach: a no-code platform and an API. This feature makes it accessible to everyone, from tech-savvy developers to domain experts and business executives—all of whom can evaluate AI applications for accuracy and consistency without needing in-depth programming knowledge.

A Smart Solution for AI Evaluation

So, how does Composo work? At its core, the technology combines a reward model trained on ideal outputs with specific criteria tailored to each application. For instance, a medical triage chatbot can define criteria to spot red-flag symptoms, and Composo scores the bot’s performance against those guidelines. This tailored approach offers precise insights into the application’s reliability.

Recently, Composo launched Composo Align, a public API that enables the evaluation of LLM applications based on any criteria. This innovation has attracted notable clients, including Accenture, Palantir, and McKinsey, alongside a recent two million-dollar pre-seed funding round. In today’s booming AI landscape, where funding is plentiful, this figure may seem modest, but according to Composo’s co-founder and CEO, Sebastian Fox, this is due to their capital-efficient strategy.

Growing the Team and Technology

Fox, a former consultant at McKinsey, emphasized that their focus is not on developing foundation models but on enhancing the usability of existing models. “If I wake up to news of a significant advancement from OpenAI, that’s great for my business,” he noted. With the fresh funding, Composo aims to hire more talent, expand its client base, and invest in research and development. "Our goal this year is to scale our technology across our partner companies," Fox added.

This latest funding round was led by British AI pre-seed fund Twin Path Ventures, with involvement from JVH Ventures and EWOR, which previously supported Composo through its accelerator program. A Twin Path spokesperson highlighted Composo’s unique positioning: “They are addressing a critical bottleneck in the adoption of enterprise AI.”

Overcoming the Bottleneck

For many enterprises, the excitement surrounding AI has subsided, replaced by pragmatic considerations about its actual impact on their operations. “Businesses are asking if AI can enhance their workflows,” Fox explained. “Is it reliable enough to mitigate reputational risks?" This skepticism provides an avenue for Composo’s offerings, especially in industries like compliance, healthcare, and security, where reliability is paramount.

Fox expressed confidence in their competitive edge, arguing that the R&D efforts behind Composo Align represent a significant barrier to entry for potential competitors. "There’s a distinct value in the architecture of our model and the expert data we’ve utilized for training," he explained.

While tech giants could easily enter the evaluation space with their resources, Composo believes it can leverage its first-mover advantage and the valuable data gathered from client evaluations. This agility positions them well as the industry shifts towards increasingly agentic AI solutions. “We’re not in a place where agents are fully effective yet, but that’s the problem we’re here to solve,” Fox asserted.

Conclusion

In a world where AI adoption can be daunting, companies like Composo are paving the way for more reliable, evaluative approaches. By focusing on a diverse range of criteria and easing accessibility, they’re addressing critical issues that the industry faces in embracing AI technology.

The AI Buzz Hub team is excited to see where these breakthroughs take us. Want to stay in the loop on all things AI? Subscribe to our newsletter or share this article with your fellow enthusiasts.