Features

Everything you need to build, deploy, and govern AI agents at scale.

Flagship

Agent Studio

The workspace where agents come to life. Agent Studio is a visual environment for building, testing, and iterating on agents — without losing the power of code. Define behaviour, connect tools, set governance policies, and watch your agent execute step by step, all in one place.

agent-studio — support-triage-v3

Agent Config

Modelclaude-sonnet-4
Tools4 connected
Policystrict
Budget$2.50/run

Execution Trace

● classify_intent 12ms

● lookup_customer 84ms

● check_policy 3ms

● draft_response running...

○ send_reply

Run Stats

Tokens2,847
Cost$0.18
Latency1.2s
Statusexecuting

Live testing

Run your agent against real inputs and watch every step in real time — tool calls, decisions, and branching paths.

Step-through debugging

Pause execution at any point. Inspect context, modify variables, and resume. Debug agents the way you debug code.

Version history

Every edit is versioned. Compare agent behaviour across changes, roll back to any previous state, and branch experiments.

Cost simulation

See projected costs before you deploy. The studio estimates token usage and tool calls based on your test runs.

Governance & Permissions

Role-based access, approval workflows, and granular permissions. Every agent action is policy-checked before execution.

  • Define per-agent permission boundaries
  • Require human approval for sensitive operations
  • Set execution budgets and rate limits
  • Policy-as-code with version control
[illustration]

Durable Execution

Long-running agent tasks survive restarts, network failures, and deployment updates.

  • Automatic checkpointing and recovery
  • Exactly-once execution guarantees
  • Timeout and retry policies
  • Progress tracking for multi-step workflows
[illustration]

Knowledge Management

Connect your docs, databases, and APIs. Agents retrieve the right context automatically.

  • Built-in RAG pipeline with chunking and embeddings
  • Support for PDFs, Notion, Confluence, and more
  • Automatic reindexing on source changes
  • Relevance scoring and citation tracking
[illustration]

Agent Composition

Build complex workflows by chaining specialised agents with conditional logic and branching.

  • Visual workflow builder
  • Parallel and sequential execution
  • Conditional branching and loops
  • Shared context between agents
[illustration]

Multi-Channel Deployment

Deploy agents across Slack, email, web chat, and APIs from a single definition.

  • Channel-specific message formatting
  • Unified conversation history
  • Webhook and API triggers
  • Custom UI components for web chat
[illustration]

Full Audit Trail

Every decision, tool call, and cost — logged and searchable for compliance.

  • Structured logging with search and filters
  • Cost attribution per agent, team, and workflow
  • Export to your SIEM or data warehouse
  • Retention policies and data governance
[illustration]