Composio Secures $25M to Revolutionize AI Agent Development with Experience-Based Learning

Composio raises $25M in funding to ease AI agent development - SiliconANGLE

Composio, a startup with a platform for building artificial intelligence agents, today announced that it has raised $25 million in funding. Lightspeed Venture Partners led the Series A round. It was joined by SV Angel, HubSpot Inc. founder Dharmesh Shah and more than a half dozen other backers. Composio's outside funding now stands at about $29 million. Many of the building blocks necessary to build an AI agent are available in the open-source ecosystem, which spares developers the hassle of writing everything from scratch. But the process can still take upwards of days depending on an agent's complexity. Composio, officially Sampark Inc., has developed a cloud platform that it says can compress the workflow into a few minutes. One of the most time-consuming tasks involved in building an AI agent is implementing its authentication workflow. A programming agent, for example, must authenticate itself to GitHub before it can review code stored in the Microsoft Corp.-operated service. The process must be repeated for every development team that connects the agent to its GitHub repository. According to Composio, its platform automates much of the work involved in implementing agent authentication workflows. That includes the task of managing authentication tokens, which is one of the most complicated chores involved in the process. Authentication tokens are files that have a similar function as passwords. After developers integrate their agents with Composio, they can connect them to more than 3,000 cloud applications and other software tools. The manner in which agents interact with those tools can be customized. A software team could, for example, require that an agent explain the calculations it performs using a third-party analytics tool. Developers can also connect custom software tools to their Composio-powered agents. An MCP server, in turn, makes it possible to establish application connections using the Model Context Protocol. The technology automates several of the tasks involved in integrating agents with external systems such as databases. Composio is compatible with open-source agent development frameworks such as OpenAI's Agents SDK. According to the company, those frameworks enable customers to create agents that not only respond to prompts but also carry out actions on behalf of the user. Under the hood, Composio's platform is powered by what it describes as a reinforcement learning layer. The company says that the software allows customers' AI agents to gradually improve their capabilities over time. Furthermore, agents can share information with one another on how to complete tasks. "The challenge isn't making AI smarter in isolation," said Composio Chief Executive Officer Soham Ganatra (pictured, left, with co-founder Karan Vaidya). "It's giving AI the ability to accumulate practical knowledge the way humans do -- but at the scale and speed only software can achieve."

Composio Raises $25 Million to Help AI Agents Learn from Experience | AIM

Agentic artificial intelligence (AI) startup Composio has secured $25 million in Series A funding led by Lightspeed Venture Partners, bringing its total funding to $29 million. The round also saw participation from Guillermo Rauch (Vercel CEO), Dharmesh Shah (HubSpot CTO & Founder), Gokul Rajaram, Soham Mazumdar (Rubrik Co-founder), and institutional investors SV Angel, Blitzscaling Ventures, Operator Partners, and Agent Fund. Existing investors Elevation Capital and Together Fund also joined the round. The company is focused on addressing what it sees as the core limitation preventing AI agents from transforming enterprise workflows -- the inability to learn from experience. "You can spend hundreds of hours building LLM tools, tweaking prompts, and refining instructions, but you hit a wall," said Soham Ganatra, CEO of Composio. "These models don't get better at their jobs the way a human employee would. They can't build context, learn from mistakes, or develop the subtle understanding that makes human workers invaluable. We're solving this at the infrastructure level." Composio's platform includes a shared learning layer for AI agents. This allows knowledge gained by one agent -- such as handling a Salesforce edge case or optimising a GitHub workflow -- to be reused by others. The company says this creates a network effect, enabling all agents on the platform to improve collectively. Founded two years ago by Ganatra and Karan Vaidya, Composio began building its infrastructure before the rise of current AI agent hype. The company focused on problems such as multi-agent coordination, authentication across enterprise tools, and scalability. A core innovation has been a reinforcement learning layer that allows agents to build intuition from experience. "The challenge isn't making AI smarter in isolation," Ganatra added. "It's giving AI the ability to accumulate practical knowledge the way humans do -- but at the scale and speed only software can achieve." The platform has gained traction among developers and startups. Over 100,000 developers have joined, with more than 200 companies -- including Glean -- using Composio. Early adopters include startups from recent Y Combinator batches such as April, OpenNote, Airweave, Den, and Dash. AI products like Context and Altera have also built on the platform. A common use case is reducing the time it takes to build and ship agents. Instead of creating new integrations from scratch, developers can access pre-built capabilities from Composio's library of learned behaviours. "What excites us about Composio is that they're not just solving today's integration problems," said Raviraj Jain of Lightspeed Venture Partners. "They're building the foundation for AI agents to become genuinely useful by learning from experience at scale. This is the missing piece between impressive demos and transformative deployments." The funding will be used to further develop Composio's learning infrastructure. The platform integrates with frameworks like MCP, LangChain, Vercel AI SDK, and OpenAI Agents, allowing developers to adopt its learning capabilities across different tech stacks.

Agentic AI startup Composio raises $25 million in funding round led by Lightspeed Venture Partners - The Economic Times

Existing investors Elevation Capital and Girish Mathrubootham's Together Fund, besides angel investors such as Gokul Rajaram, Rubrik cofounder Sohum Mazumdar, HubSpot founder Dharmesh Shah, and others, participated. The company will use the funds to expand its engineering and research team. To date, the startup has raised a total of $29 million.Agentic artificial intelligence (AI) startup Composio has raised $25 million in funding led by Lightspeed Venture Partners, the company's cofounder and CEO, Soham Ganatra, told ET in an interaction. The round also saw participation from existing investors, such as Elevation Capital and Girish Mathrubootham's Together Fund, in addition to angel investors, such as Gokul Rajaram, Rubrik cofounder Sohum Mazumdar, HubSpot founder Dharmesh Shah, and others. The company had earlier raised $4 million in seed money, taking its total fundraise to $29 million. Ganatra said that the company will use the funds to expand its engineering and research team. Composio simplifies complex enterprise workflows through AI-driven automation and has over 200 companies as paying customers. It is generating over $1 million in annualised recurring revenue, Ganatra said, without disclosing specific revenue details. The startup, based in San Francisco, with a development centre in Bengaluru, plans to increase its team size from 25 currently to 40 by the end of this year. Composio is building infrastructure that lets AI agents plug directly into widely used business apps like Gmail, GitHub, Salesforce, Slack, and others. It acts as a connective layer between autonomous AI tools and the enterprise software stack. Its platform offers pre-built, production-ready integrations, allowing AI agents to perform actions like sending and organising emails, updating customer relationship management (CRM) entries, managing tickets, and even interacting with code repositories -- without developers needing to build each connection from scratch, deal with complicated logins, or write and maintain extra code. This reduces the friction of deploying AI in real-world business environments, where legacy systems, security constraints, and integration overhead often slow things down. "You can spend hundreds of hours building LLM tools, tweaking prompts, and refining instructions, but you hit a wall," Ganatra said. "These models don't get better at their jobs the way a human employee would. They can't build context, learn from mistakes, or develop the subtle understanding that makes human workers invaluable. We're solving this at the infrastructure level." Ganatra said that over 100,000 developers use the platform, with adoption gathering pace among AI-first companies. Top startups from the latest Y Combinator batches like April, OpenNote, Airweave, Den, and Dash are Composio's customers, he said. Composio's funding comes at a time when risk investors are increasingly backing cross-border startups building AI solutions. Healthcare-focussed startup Risa Labs, which aims to improve cancer care workflows, raised $3.5 million in a seed funding round back in April, led by Flipkart cofounder Binny Bansal with participation from General Catalyst, z21 Ventures, and others. In January, Khosla Ventures and Z47 (formerly Matrix Partners India) led a $25 million funding round for Atomicwork, which is building AI agents to help enterprises manage their IT workflows. In December last year, task automation venture RapidCanvas raised $16 million in a round led by Peak XV Partners.