Description
Enterprises have spent billions of dollars integrating REST APIs to infuse business logic, workflows, and data into their application portfolios; they’re expected to continue spending to the tune of $30 Billion by 2033 according to some estimates, with the expectation that these investments will reinforce their aspirations for AI. As AI agents enter production systems, a new challenge appears: how do you make those existing APIs safely usable by autonomous agents without rewriting them?
Most teams start by exposing APIs as tools using MCP (Model Context Protocol). This is a pragmatic first step. MCP provides a standardized execution surface and gives agents access to enterprise capabilities. But it mainly answers what an agent can do. Workflow, sequencing, and constraints often remain outside the API, embedded in prompts or agent logic.
Please join Adib Saikali as he introduces a complementary approach to API integration based on HATEOAS (Hypermedia as the Engine of Application State). By allowing APIs to describe allowed actions and transitions at runtime, hypermedia shifts control back to the server—where enterprises already enforce policy, validation, and governance.
In this session we will compare MCP and hypermedia head to head and show why the strongest agent architectures combine both MCP to expose capabilities, and hypermedia to guide and constrain agent behavior—unlocking existing APIs for AI at enterprise scale.
Speakers
Adib Saikali, Distinguished Software Engineer, Broadcom
Wednesday, March 25, 2026
On Demand
