At Sampark Inc, higher identified for its platform, Composio, co-founders Soham Ganatra and Karan Vaidya are dragging the idea to the AI period. They’re providing a developer-focussed integration platform constructed particularly for AI brokers and enormous language fashions (LLMs)—enabling AI programs to speak to one another and the true world.
“Fashionable software program at the moment is like form of compound merchandise. They require knowledge pulling in from completely different sources on the market,” Ganatra tells Forbes India. “So, integrations are a needed a part of product constructing. It’s so integral that with out it, the product truly can’t exist.”
Composio presents a platform the place AI brokers and LLMs hook up with a whole lot of instruments, from assets akin to GitHub to HubSpot, with only a line of code. It’s a ‘developer-first’ toolset that cuts integration time from months to days and improves reliability of AI brokers from round 50 % to as a lot as 90 %, the entrepreneurs declare.
Ganatra and Vaidya, engineering graduates from IIT-Bombay, have identified one another from their days collectively as physics Olympiad teammates in 2013. They based Composio (Sampark) in 2023, based mostly in San Francisco and Bengaluru.
Composio is what software program product specialists akin to Nate Jones name ‘middleware’. “The premise is easy: Let builders concentrate on what the AI ought to do, and let Composio deal with how to connect with the apps and knowledge that allow these actions,” Jones explains in a March 8 version of his well-liked e-newsletter on Substack.
The platform can be an instance of an rising tech known as mannequin context protocol, based on Jones. The standard path is daunting if one desires to construct an AI assistant to deal with mail, a platform like Slack, a buyer administration software program and a database all on the identical time. There are a number of duties which might be tedious and may take months, he writes.
“Composio acts as a unifying layer between AI brokers (like LLMs) and the online of third-party providers they should use,” he explains. It presents options akin to pre-built integrations, managed authentication, and higher execution, which reduces issues akin to LLM hallucinations.
View the full list here
“So for a brand new developer, it’s the hassle of going dwell that differs. There’s simply a 95 % discount of their effort,” Ganatra says. “We’ve got seen tasks the place they thought it could take them 4 to 5 months to completion. We’ve got actually seen them go dwell in every week.”
“One of many greatest challenges in agentic AI is integration—AI brokers must seamlessly work together with enterprise functions, instruments, and APIs with out human intervention. That’s the place Composio is available in,” Manav Garg, founding managing accomplice at Collectively, a Bengaluru VC fund, and an early backer of the startup, writes in a February put up on LinkedIn.
Additionally learn: Niki Parmar: Pushing the boundaries of AI
“In some ways, Composio is to agentic AI what Twilio was to communication APIs: A foundational layer that can unlock a whole trade,” Garg provides. Because of this over 14,500 builders, together with groups from Meta, Salesforce, and Cisco, have adopted Composio’s framework, he says.
Right now that quantity has greater than doubled, based on Ganatra. “We’re rising at greater than 50 % month-on-month when it comes to simply customers,” he says. “By way of income, we’re in seven figures proper now,”.
In March, Collectively teamed up with Lightspeed Enterprise Companions and Elevation Capital for a $24 million Sequence A funding in Composio.
Wanting forward, a prime precedence is discovering the suitable folks for the suitable roles. On the product entrance, “the place we’re going is attempting to create a brand new layer for AI brokers”, he says.
This implies, as AI brokers and their interactions with a number of software program and providers develop into extra widespread, that layer should enable “any type of intelligence to study from expertise, to create and to have instinct”. With extra difficult operations, can the agent get higher over time, he asks.
(This story seems within the 13 June, 2025 challenge
of Forbes India. To go to our Archives, click here.)