Embedded vs. AI Agent Studio
Two paths to adopting AI agents — Oracle's quarterly-shipped embedded agents, or partner/customer-built agents via AI Agent Studio. Most customers use both.
Path 1 — Embedded agents (the default)
Oracle ships AI agents as part of the quarterly Fusion release. You enable them like any other feature: in the readiness portal, you flag the agents you want, then turn them on after the upgrade.
Use this path for:
- Common workflows Oracle has already built for (invoice processing, deal advice, talent review, etc.).
- Anything where "out of the box" gets you 80% of the value.
- Most customers, most of the time.
Tradeoff: you adopt at Oracle's pace. If a workflow you care about isn't covered yet, you wait.
Path 2 — AI Agent Studio (since Release 25C)
For agents Oracle hasn't built, AI Agent Studio is a Fusion-native, no-code platform to configure and deploy your own. It includes:
- Agent templates to start from.
- No-code configuration of agent goals, tools, and data access.
- Native Fusion integration — same security, same audit, same data model.
- Third-party tool integration for actions outside Fusion.
Use this path for:
- Industry-specific workflows.
- Custom internal processes.
- Bridging Fusion to your other systems.
Path 3 (kind of) — AI Agent Marketplace
Oracle launched a Marketplace in October 2025 where partners publish validated, ready-to-deploy agents. Think of it as the App Store for Fusion AI: you browse, install, configure, run.
How most customers actually do this
- Adopt all relevant embedded agents quarterly (low cost, high value).
- Use the Marketplace to fill industry-specific or partner-built gaps.
- Build custom agents in AI Agent Studio only when (1) and (2) don't cover the need.
In that order. Don't start with custom — you'll over-invest in scaffolding Oracle is going to ship anyway.
Action checklist
Tap each step as you complete it.