Talent · Enterprise AI

General Motors:
AI agents live across five departments.

Five enterprise departments were running critical workflows through disconnected vendors and manual work. Incipient placed an embedded Enterprise AI SME who deployed purpose-built Glean agents wired to live enterprise data, retiring external vendor contracts and putting AI where the work actually happens.

5
Departments with AI agents live
2
Vendor contracts eliminated
MCP
Live enterprise-data access
GENERAL MOTORS · AGENTS LIVE BY DEPARTMENTLegalagents liveScreening agentMarketing agentFinanceagents livePurpose-builtagents on live dataHRagents livePurpose-builtagents on live dataOperationsagents livePurpose-builtagents on live dataMarketingagents livePurpose-builtagents on live dataLegal retired two external vendor contracts: a screening vendor and a content vendor.All five departments draw on live enterprise data through MCP, governed from launch.
Where the agents are live. Legal went furthest, replacing two external vendors; the other departments run purpose-built agents on live enterprise data.
Industry
Automotive, Fortune 500
Focus
Enterprise AI
Model
Embedded AI SME
Stack
Glean, MCP, Web SDK
Read moreRead less
The Challenge

Critical workflows, disconnected vendors, no AI layer.

Legal, Finance, HR, Operations and Marketing were running critical workflows through external vendors and manual processes, paying recurring fees for contract screening, legal content and data work.

There was no AI layer, no integration to internal systems and no governance at scale. The enterprise needed purpose-built agents connected to live internal data, embedded in the tools people already use, and governed across five departments at once.

  • 1Recurring vendor feesContract screening and legal content were bought from external vendors.
  • 2No AI layerNone of the five departments had AI in their workflow.
  • 3No internal integrationNothing connected to internal databases or document repositories.
  • 4No governance at scaleAccess, security and connectors had no owner across departments.
  • 5Manual workScreening, content and data work was done by hand.
What We Did

Purpose-built agents, wired to live enterprise data.

Incipient placed a senior Enterprise AI SME embedded inside the AI platform team, which meant architectural ownership inside the enterprise rather than at arm's length.

  • Built a Legal Screening Agent for contract review, risk-clause flagging and precedent surfacing, replacing an active third-party screening vendor
  • Built a Legal Marketing Agent for communications and collateral generation, eliminating a separate content vendor
  • Gave the agents real-time programmatic access to internal databases and document repositories through MCP connectors
  • Put the agents inside existing enterprise applications with Web SDK embeddings and API and OAuth integrations, reading and writing back
  • Managed access, security and connector governance across every department
GENERAL MOTORS · HOW AN AGENT WORKSEmployeeexisting appAgentGleanMCPconnectorsDataDBs + documentsWrite backAPI / OAuthWeb SDKembedcallsMCPreadslivereturnsresultevery agentGovernanceaccess/securityBEFOREdisconnected vendors and manual work · no AI layer · no governanceAFTERagents live in 5 departments · 2 vendor contracts retired
How an agent sits in the work: embedded in an existing app, reading live enterprise data through MCP, writing back, and governed for access and security. Orange marks the agent and its data path.
The Outcome

AI agents live across five departments.

The Legal department retired multiple external vendor contracts outright. Purpose-built agents now run across Legal, Finance, HR, Operations and Marketing, drawing on live enterprise data through MCP and surfaced inside the tools employees already use.

They were governed for access and security from launch, and the deployments in production are the foundation for continued expansion.

5
Departments with AI agents live
2
Vendor contracts eliminated
MCP
Live enterprise-data access
Verified Reference
"They transformed how our enterprise uses AI, not by deploying a tool, but by redesigning entire workflows around it."
Enterprise AI Lead · Fortune 500 automotive
Clutch verified reference →
Ready to put AI where the work actually happens?
Tell us what's holding you back, and we'll show you what it's costing you and fix it.