Software developers’ time is highly valuable because their role is critical to the company’s operations.
Compositio, which was painted in June your story Tech30 2024 list, builds tools for developers. Its platform accelerates the process of building and connecting agents to external services, reducing development time from months to days.
The San Francisco, California-based startup offers tools that enable developers to quickly create reliable AI agents that interact with external software such as CRM systems (eg, Salesforce, HubSpot) and services such as Gmail.
These AI agents are essentially automated systems that perform tasks on behalf of the user on those software platforms. For example, an agent can be designed to automatically manage emails, update records in CRM, or perform data entry without human intervention.
Typically, when developers build such agents, it can take months to fine-tune them to work correctly and reliably.
Composio simplifies agentic integration—how multiple AI agents communicate and act together—by providing a platform where developers can connect their agents to these systems more quickly and reliably. Its platform has gained traction on GitHub with over 12,000 developers building on its framework, including engineers from Fortune 500 companies such as Meta Platform, Salesforce and Cisco.
“AI agents are inherently unreliable and prone to confusion, making them one of the biggest challenges to build. When agent integrations go live in production, they often operate with significant inaccuracies, requiring developers to spend months optimizing reliability,” Compozio’s co. – Says Founder Soham Ganatra. your story.
The startup was launched in June 2023 by IIT-Bombay alumni, Karan Vaidya and Soham Ganatra.
A one-size-fits-all solution
Most AI agents or LLMs rely on the ability of an AI model (such as an LLM) to interact with external functions or tools to perform specific tasks, essentially called tool calls.
However, the reliability of function calling is often very low due to confusion and inaccuracy–the main reason why many LLM applications are not widely used in production.
For example, when you go directly with an agent interacting with Gmail, the reliability will be somewhere closer to 40% to 50%. According to Ganatra, Composio’s reliability starts above 90 percent, which will allow developers to go into production quickly.
Composio’s solutions work at a fundamental level and apply agent integration across industries, improving AI integration regardless of specific use cases.
Pricing and customers
Like many other AI applications, Composio operates on a hosted platform with a usage-based pricing model. Developers and large enterprises can sign up, create an account and access Composio’s services using an API (Application Programming Interface) key.
They access Composio’s services through an API (Application Programming Interface) key, and pay fees each time the API is called.
This pay-per-use approach enables flexibility for different scales of operations and promotes greater adoption. Its average contract price is currently close to $20,000 or Rs 16 lakh.
Its customers and partners include enterprises such as startups like Databricks, Datastax or 11x.
It is used by developers, who naturally represent enterprises, and monetization on that front has also started less than a month ago.
“The AI ​​agents space in general is very nascent and so the goal is not yet monetization, but the goal is actually being used,” Ganatra explains.
About 80-90% of projects involving agents and LLMs are currently in the MVP (Minimum Viable Product) or POC (Proof of Concept) stage. When these projects move into full-scale production, that is expected to become more mainstream around February or March 2025, adds Ganatra.
Composio is competing for a small slice of the enterprise AI market that is expected to reach $311.64 billion by 2029, growing at a CAGR of 52.17%, according to Mordor Intelligence.
As AI adoption accelerates, seamless integration and seamless workflows are critical for enterprises to remain competitive. By shortening the development cycle, it enables businesses to take full advantage of AI’s potential while meeting the demands of a rapidly evolving industry, enabling time-to-market, lower costs, and focus on innovation.