Guiding Principles — Workato Labs

01 Agent-first

Built for agent-assisted development.

The way developers build integrations is changing — we build for that reality first.

AI coding agents are becoming primary collaborators — not just autocomplete, but partners that read documentation, generate code, validate output, and iterate. Every tool we ship assumes an agent is in the loop: skills are structured as agent-consumable knowledge, the linter provides deterministic validation that agents can't do themselves, the CLI exposes a workflow agents can drive end-to-end.

We design for how developers will work next year, not how they worked last year.

02 Open

Open by default.

When we have to choose between optimizing for one tool and staying portable, we stay portable.

Everything we ship is built on open standards and designed so developers can use it with whatever tools they already have — any AI agent, any IDE, any OS. No lock-in to a specific vendor, framework, or Workato-proprietary format. Open source licensing, open contribution model, open roadmap.

Tight integrations with specific tools — like Claude Code slash commands — ship as documented bonuses on top of the open layer, never as the assumed entrypoint.

03 Craft

Developer experience is a first-class feature.

Not just “it works” — it's pleasant to use from day one.

One-command install. Clear error messages. Documentation that respects your time. A workflow that composes naturally without requiring you to understand the architecture underneath. We hold every tool in the Workato Labs portfolio to this standard, not just the flagship ones.

If a developer's first five minutes with any of these tools feels rough, that's a bug.

04 Contribute

Easy to contribute — for everyone.

The contribution surface is as deliberately designed as the user-facing experience.

Repo structure, documentation, and tooling should invite contributions from the full spectrum of people building with Workato: experienced developers, vibe coders experimenting with AI-assisted workflows, and the agents themselves.

Adding a new connector skill shouldn't require understanding the CLI codebase. Filing useful feedback shouldn't require writing Go.

05 Velocity

Ship fast, learn in the open.

The experimental label is an operating philosophy, not just a disclaimer.

We ship early, gather real usage signal, and iterate. No SLAs, no heavyweight release gates. Community-reported gaps flow directly into connector and platform roadmaps — this is a product intelligence channel, not a side project.

Core engineering teams use the same tools and contribute back, so the ecosystem grows from the inside and the outside simultaneously.