Pricing
$699 per agent.
One price. One model.
Every agent is its own machine. Own process, own memory, own audit trail. Scale by adding agents, not feature tiers.
Why per-agent
Most AI tools charge per seat. We charge per agent because one agent is one autonomous system.
Each agent runs in an isolated process with its own memory, its own invariants, and its own append-only audit log. Two agents under the same company never read each other's data unless you explicitly wire a shared channel. An agent scoped to your AUSTRAC workflow cannot leak into the one scoped to client FX reporting.
This is the unit of isolation, the unit of compliance, and the unit of price. Adding a user who talks to an existing agent is free. Adding a new autonomous system costs $699/mo.
How teams scale
Three shapes, same per-agent price.
Solo
One agent. One operator.
$699/mo
1 agent · billed monthly
- ✓Agent running in 60 seconds · full onboarding (13 steps, 5 phases) in one session
- ✓48-hour sandbox (subject to fair-use limits). Runs on our API until you paste your key
- ✓BYOK (Claude, GPT, Gemini), or managed Claude subscription as convenience opt-in.
- ✓Full audit export on request
Most common
Team
Two to five agents. One shared policy stack.
$699/agent/mo
Team of 5 · $2,199/mo bundled
- ✓Everything in Solo
- ✓Guided onboarding call · 30 min with Isaac
- ✓Shared invariant library across agents
- ✓Opt-in cross-agent memory channel
- ✓Priority review when invariants fire
Scale
Five or more agents. Regulated deployment.
Let's talk
5+ agents · volume + compliance terms
- ✓Everything in Team
- ✓Custom compliance framework (beyond AUSTRAC)
- ✓Dedicated review cadence with adversarial reviewer
- ✓Volume pricing on the per-agent rate
- ✓DPA, custom data residency, on-prem option
Included with every agent
The same governance floor. No feature paywall.
- ✓Pattern-to-invariant learning pipeline
- ✓Adversarial review gate before ship
- ✓Behavior monitoring + audit trail export
- ✓Human-in-the-loop escalation
- ✓Governance hooks fire before every action
- ✓7-year audit retention, aligned to AML/CTF Act s.116(3) retention clock
- ✓Isolated tenant storage · Sydney region
- ✓Free intake assessment · no credit card
Questions we keep getting asked
FAQ.
Why not per-seat like everyone else?
Per-seat rewards vendors for adding more humans to the tool. Per-agent rewards us for keeping each agent focused, isolated, and auditable. If ten humans ask the same agent the same question, the audit trail is one trail. Not ten. That's the unit you actually pay for.
What if I start solo and grow into a team?
Add agents as you need them. No re-platforming. The second agent inherits your first agent's invariant library on day one unless you explicitly fence them apart. Billing pro-rates mid-month.
Can two agents share memory?
Opt-in only, and only via a declared channel. Isolation is the default because isolation is what makes the audit trail defensible. You won't find out after a breach that agent A could read agent B's DMs.
My industry isn't AUSTRAC. Is this still for me?
The AUSTRAC framework ships by default because we built against the hardest regulator we know. The same invariant engine runs whatever framework you bring. AFSL, APRA, Privacy Act, REA disclosure rules, medical records, your internal compliance policy. Scale tier includes custom framework authoring.
Is there a free tier?
No. The first 48 hours run on our API key (subject to fair-use limits) so you can see the agent working before you commit. After that it's $699/mo per agent, monthly, cancel anytime. No credit card to start the intake.
What happens if I cancel?
Agent stops on the day the billing period ends. Full JSONL audit export is sent to the admin email within 24 hours. Tenant data is purged after 30 days unless you've requested retention for regulatory reasons.
Intake takes three minutes.
Tell us what you want the agent to do, what the worst mistake would be, and where to reach you. Agent running in 60 seconds; full onboarding (13 steps, 5 phases) in one session.
No credit card. Free consultation included.