MIT Sloan on "Agentic AI" Business Models
MIT Sloan is promoting a framework for how business models will evolve in the era of agentic AI. The key shift is from selling software or services to selling outcome-oriented autonomous systems that perform tasks on a user's behalf.
The MIT Sloan framework identifies four business models for the agentic AI era: "Existing+," which augments current models with AI; "Customer Proxy," where AI executes predefined processes to achieve customer outcomes; "Modular Creator," using AI to assemble reusable modules for outcomes without a fixed process; and "Orchestrator," where AI assembles a whole ecosystem of products and services. This evolution builds on a 2013 framework, responding to a market where "ecosystem driver" business models grew from 12% of businesses in 2013 to 58% in 2025. The economic promise of agentic AI lies in its ability to dramatically lower transaction costs—the effort involved in searching, communicating, and contracting. This is attracting significant venture capital, with funding for agentic AI startups nearly tripling to $3.8 billion in 2024 from $1.3 billion in 2023. Projections indicate that agentic AI could account for 10% of all AI funding in 2025, representing about $6.7 billion in investment. In India, there are over 133 companies in the agentic AI space, which have collectively raised $290 million. Notable funded startups include Avataar, Sarvam, and Atomicwork. The Indian IT giants are also heavily involved, with companies like Cognizant, Infosys, TCS, and Wipro collectively deploying over 200,000 Microsoft Copilot licenses to embed agentic AI into their workflows. For developers, the rise of agentic AI presents a new frontier. Bangalore-based Composio, founded by Soham Ganatra and Karan Vaidya, is building a developer-centric integration platform to help AI agents connect with external applications, a critical challenge in the space. Their framework is already in use by over 14,500 developers at companies like Meta and Salesforce. Another Indian startup, Emergent, founded by Mukund and Madhav Jha, is creating an AI coding agent to automate the entire software development lifecycle. On Hacker News, developers discuss the practicalities of building with agentic AI, framing agents as "junior team members" that require clear scope and explicit success criteria. The consensus is that the real value isn't just in writing code faster, but in orchestrating complex tasks and applying domain expertise—a sentiment echoed by CTOs of vertical SaaS companies who find their deep domain knowledge is not easily replicated by internal developer teams using AI agents. The shift is from selling tools to selling outcomes. Instead of a developer buying a license for a CI/CD tool, they might subscribe to an agentic system that guarantees a certain level of code quality and deployment frequency. This requires a focus on reliability and observability; as one developer noted, integrating tracing with tools like Grafana can be a significant challenge, but it's crucial for debugging and optimization. Gartner forecasts that by 2026, 40% of enterprise applications will have integrated task-specific AI agents, up from less than 5% today. This creates opportunities for founders to build "smart middleware" that connects modern agentic interfaces with legacy systems, or to create new, agent-native solutions for specific verticals. The key will be to balance AI autonomy with human oversight and to solve the "sociotechnical aspects" of implementation, which research suggests can take four times the effort of perfecting the AI model itself.
