Agent Studio: from two products to one mental model
Consolidating two separate workspaces into one platform — and letting brutal research findings steer every turn.
- Role
- VP of Product Design & Research
- Team
- Design + research across Platform, AI, Security, UI
- Timeframe
- 2024–2026
- Outcome
- Unified platform; AI Hub, test mode, enterprise parity shipped
Two products, one confused user
Workato's agent capabilities lived in two separate workspaces with two mental models. Users didn't experience two products; they experienced one product that disagreed with itself. We consolidated them into Agent Studio — launched June 2025 — coordinating design across Platform, AI, Security, and UI teams.
The finding we didn't want
Then we ran the research that hurt. A builder study with a major observability vendor [CLEARANCE] naming Datadog found us to be, in their words, the most complicated agentic platform on the market. They dropped us from a proof-of-concept because their non-technical stakeholders couldn't participate. Build times ran three weeks against a market expectation of one. Our terminology — genies versus agents, skills versus tools — confused even experienced users.
Turning the ship
That finding set the roadmap. Disintegrated experiences became in-context ones: agent creation opened inside the recipe editor where the work already lived, instead of a separate destination. We shipped AI Hub, test mode, and enterprise parity — BYOLLM, knowledge-base management, conversation RBAC — while cutting the concepts a builder had to hold in their head. [VERIFY] build-time improvement figure.
I didn't do most of this. My entire team did. My job was making sure the hardest research finding outranked the org chart.