Session

Provisioning for the Agentic Era: How Databricks Built a Self-Serve Infrastructure Vending Machine

Overview

ExperienceIn Person
TrackApplication Development
IndustryCommunications, Media & Entertainment
TechnologiesDatabricks Apps
Skill LevelIntermediate

As Databricks' go-to-market organization grew past 7,000 people — and AI agents became first-class consumers of platform resources alongside human engineers — our field team ran into a familiar problem: shared demo environments couldn't keep up. Too many users needed admin-level access, platform limits became real constraints, and it was hard to see who was using what, whether that "who" was a person or an agent.

 

In this session, we'll share how we addressed that by building the Field Engineering Vending Machine (FEVM), a Databricks App that gives every engineer — and every agent — just-in-time, isolated, and governed environments on demand. Instead of filing tickets or juggling shared workspaces, engineers describe what they're trying to do — build a demo, run a hackathon, reproduce an issue, spin up an agentic workflow — and FEVM delivers a fit-for-purpose environment in minutes, then cleans it up automatically when it's no longer needed.

 

We'll show how this "vending machine" model has improved speed, control, and cost visibility for thousands of users and countless agent runs, and why building it as a Databricks App has become a blueprint for how we dogfood the platform internally — for the agentic era and beyond.

Session Speakers

Speaker placeholderIMAGE COMING SOON

Greg Wood

/Lead SSA
Databricks

Joel Thomas

/Director
Databricks

Full Summary

Inside Databricks' FE Vending Machine: how an internal team built an agentic self-serve infrastructure platform

Databricks' field engineering organization outgrew its shared demo workspaces as the company scaled from hundreds to thousands of go-to-market users. The result was fragile demos, mounting cost and security risks, and an overwhelmed admin crew. An internal tooling team responded by building the FE Vending Machine (FEVM), a self-service, agent-driven system that provisions just-in-time, governed infrastructure on demand, then reclaims it automatically.

FAQ


FEVM is an internal self-service system that provisions ephemeral Databricks workspaces and related resources on demand. Its primary users are solutions architects and other go-to-market staff, along with admins who operate the platform and, increasingly, agents acting on users' behalf.

Shared environments broke under thousands of users, creating compounding technical debt, security exposure, broken demos, and long request queues. Per-user, just-in-time provisioning addressed the structural limits that more capacity in a single environment could not fix.

Policies from security, legal, IT, and finance are encoded in Terraform templates and provisioning logic. Users follow the only available paths, which are already compliant, so policy adherence is automatic rather than manual.

Agents are both users and operators. An MCP-backed agent interface powers the generative UI, handles provisioning requests, and lets admins inspect deployments, approve extensions, and troubleshoot, all from chat or any MCP-accessible client.

Not necessarily. The speakers emphasize adopting the principles, not the entire implementation. For many organizations, the wins come from ephemeral provisioning, embedded governance, and agent-native workflows, even without FEVM's full scale.