Conversational Everything
Configure complex system prompts and tools entirely through natural conversation — no YAML, no JSON, no schemas to memorise.
Back to Features
v1.0 is now available
Shape the intelligence that powers your products.
The command centre for AI agents. Define templates through conversation, set behavioural rules, and deploy instances directly into your applications.
The pitch
The conversation is the product.
No complex configuration files needed. Talk to the assistant; it builds the agent with you, on the spot.
Live Visual Feedback
Watch the agent's configuration update in real time as you talk to it. The chat is the interface; the diagram is the proof.
Bring Your Own Keys
Use your own API keys for complete control over billing and data privacy. We orchestrate; the model providers serve.
How it works
From idea to production in minutes.
01
Define
Chat with the assistant to define your agent's persona, tools, and behaviour. The conversation is the configuration.
02
Deploy
Publish your template and provision isolated instances for your users via the API. One template, many deployments.
03
Iterate
Monitor instances, update templates, and instantly propagate changes to thousands of running agents.
And more
Everything you need to scale agents.
Multi-Provider Support
Seamlessly switch between Anthropic, OpenAI, and open-source models without rewriting a thing.
Instant Propagation
Update a template and automatically deploy changes to thousands of instances within seconds.
Team Workspaces
Collaborate to review, test, and publish agent templates. Versioned, auditable, undoable.
Pricing
Simple, transparent pricing.
Starter
NZD $5/month
Perfect for hobbyists and side projects.
- 3 Templates
- 100 Instances
- Community support
Team
NZD $15/month
For small teams building production apps.
- Unlimited templates
- 5,000 instances
- Priority support
- Team workspaces
Dev
PAYG
Scale instantly with pay-as-you-go pricing.
- Everything in Team
- Unlimited instances
- Custom SLA
- Dedicated account manager
Get started
Build your first agent.
Drop your email and we'll send the workspace invite plus your sandbox API key.
Frequently asked questions.
What's an instance vs a template?
A template is the agent's blueprint — system prompt, tools, model preferences. An instance is a live agent running for one of your users, derived from a template. Update the template; every instance picks up the change.
Do I bring my own model keys?
Yes. Process & Reasoning orchestrates the agent layer; the actual generations go through your Anthropic, OpenAI, or other provider keys. You keep full control over billing and data flow.
Can I self-host?
The hosted offering is the supported path today. Self-hosting is on the roadmap for Dev / Business customers — get in touch if it's a hard requirement and we'll work with you.
How do template updates propagate?
Within seconds. Publishing a new template version flips a flag the instances poll on next request, so the very next message a user sends picks up the new behaviour.
How do I integrate it into my app?
Provision instances via the REST API and route your users' messages through the instance endpoint. SDKs for TypeScript and Python ship in beta with v1.0.