Challenge 1: Automation was in place, but agents couldn't reach it
Challenge 2: No production-ready path to Salesforce or enrichment data
Challenge 3: Agent workflows hitting LLM limits at scale

People.ai had built a real automation foundation on Workato — 350+ recipes powering GTM workflows across lead engagement, sales follow-up, and operations. But as AI became a strategic priority, the team ran into a problem that many automation-first companies encounter: agents couldn't reach the systems those workflows already touched.

Salesforce's native MCP capabilities were still in development, limited to dev org environments and unavailable for production data. Waiting for vendor support wasn't an option — it would have stalled AI initiatives entirely. At the same time, People.ai wanted agents to query live CRM data alongside external enrichment sources like Apollo.io, without building and maintaining a custom integration layer for each system.

There was also a performance challenge. As agents began querying larger Salesforce datasets, early approaches that returned full object payloads — all fields, all records — created responses that pushed against LLM token limits. The team needed a way for agents to ask more precise questions of Salesforce data: specific fields, specific objects, specific filters. Without that precision built into how agents queried the system, scaling agentic workflows reliably wasn't possible.

The automation foundation was there. The question was how to make it agent-ready without starting over.

Solution 1: Deployed Workato's Salesforce Sales Cloud Explorer MCP Server
Solution 2: Wrapped Apollo.io through Workato Enterprise MCP
Solution 3: Turned 350+ recipes into Enterprise Skills agents could invoke directly

Rather than rebuilding from scratch, People.ai used Workato Enterprise MCP and pre-built MCP servers to extend their existing automation layer into one that agents could actually act on.
For Salesforce, Workato's Salesforce Sales Explorer MCP Server provided the production-ready bridge the team needed. With tools for searching leads, accounts, contacts, and opportunities through natural language, agents could query production CRM data reliably from day one. For cases requiring greater precision — specific fields, custom filters, targeted object queries — agents could use the execute_soql_query tool to construct exact queries against Salesforce data, avoiding the token bloat that comes with returning full object payloads. Before running those queries, the retrieve_semantic_model tool gave agents the schema context they needed to build them correctly. The precision that People.ai had previously needed to engineer manually was now achievable through how agents composed and directed their queries — not hardcoded into the system.
For external enrichment, the team wrapped Apollo.io through Workato Enterprise MCP, enabling agents to move between CRM and enrichment data in a single, governed workflow — without waiting for Apollo to ship native MCP support.

Because Workato served as the common MCP abstraction layer, both systems participated in the same agentic workflows through a unified access and governance model. The 350+ recipes People.ai had already built didn't become obsolete — they became the proven business logic that agents could now invoke directly, turning an automation investment into an agentic one.

With Workato powering both the automation foundation and the MCP layer on top of it, People.ai unlocked production-grade agentic AI across their GTM ecosystem — without abandoning what they'd already built.

The Salesforce Sales Cloud Explorer MCP Server enabled secure, scoped production data querying that agents could rely on in real workflows. Apollo.io enrichment extended the agentic layer across people and company records. And because everything ran through Workato's MCP gateway, every agent action was governed, auditable, and consistent with the security posture already in place. AI-driven email workflows, now informed by real-time CRM context agents could actually read and act on, drove a 20% lift in open rates.

More broadly, MCP created a second growth engine for Workato within People.ai. The automation recipes that had been running GTM workflows became Enterprise Skills agents could call on demand. Each new MCP server added capability that every existing workflow could draw on — compounding the value of the automation investment rather than replacing it.