Control plane for Arcade pipelines

Deploy Arcade agents into production with governance built in.

Arcade gives you 7,000 managed tools. AIG gives you the orchestration, approval, and audit layer companies need to actually run agents on top of them — without inventing governance from scratch.

What you get

Everything you'd otherwise build yourself.

Intent graph execution

Agents propose multi-tool plans. Humans constrain, repair, and approve before Arcade executes.

Workspace-scoped connections

Each team connects Gmail, Calendar, GitHub, Slack via Arcade OAuth — scoped per workspace, not per user guesswork.

Versioned pipelines

Save working agent workflows as named templates with parameters. Run manually or on a schedule.

Approval policies

Require reviewer roles before risky tool calls execute. Configurable per tool pattern.

Immutable audit log

Every mutation, approval, and Arcade execution is recorded. Replay any run.

Locked objectives

Repair the plan all you want — the objective is sealed at formation time and never silently rewritten.

Primary artifact

The negotiation is the artifact.

Every intent carries a co-authorship trace — agent proposals, human edits, constrained repairs, and Arcade execution results. Your team sees exactly what happened and why.