Pricing

Priced per API call. Start with 10,000 free.

Every plan includes all five live agents. What changes as you move up is the model answering your calls and the volume included — not which agents you may use.

Free trial

Free

No card required

Evaluate every live agent against your own data before you talk to anybody.

Model tier
Essential
Included calls
10,000one-off allowance
Extra calls
  • All five live agents
  • 10,000 API calls, once
  • Full audit trail and approval gates
  • One trial per company domain
  • Stops at the allowance — no bill, no surprise
Start free

Starter

$149/mo

per month

One workflow in production, at a volume that pays for itself.

Model tier
Essential
Included calls
25,000per month
Extra calls
$8/1,000
  • All five live agents
  • 25,000 calls a month
  • $8 per additional 1,000 calls
  • Mock and live connectors
  • Email support
Start free, then upgrade
Most teams start here

Growth

$899/mo

per month

Production volume on the Professional tier, with evaluation running against your cases.

Model tier
Professional
Included calls
30,000per month
Extra calls
$38/1,000
  • Everything in Starter
  • Professional model tier
  • 30,000 calls a month
  • $38 per additional 1,000 calls
  • Evaluation against your golden cases
  • Integration support during onboarding
Talk to sales

Scale

$3,499/mo

per month

The hardest reasoning, with the headroom to run it across the business.

Model tier
Advanced
Included calls
50,000per month
Extra calls
$95/1,000
  • Everything in Growth
  • Advanced model tier
  • 50,000 calls a month
  • $95 per additional 1,000 calls
  • Custom workflows and connectors
  • Priority support
Talk to sales

Prices in USD, billed monthly in arrears against the usage your portal shows. Running your own deployment? Talk to us.

Model tiers

What you are paying more for

A tier is which model answers your calls. Moving between them is a plan change, not an integration change — your key, your endpoints and your code stay exactly as they are.

Tier 1

Essential

High volume, well-defined work — classification, extraction and routing, answered fast.

Tier 2

Professional

The default for production. Reasons before it answers, across all five agents.

Tier 3

Advanced

The hardest reasoning — ambiguous documents, long multi-step operations, tight accuracy bars.

Free trial

10,000 calls, every agent, no card

Enough volume to run a real pilot against your own data, on all five live agents. It stops at the allowance rather than billing you, so there is no way to be surprised.

  • One trial per companyWe check the email domain and send a link to confirm the address is yours. Personal and disposable mailboxes are refused — the allowance is real compute, and this is what keeps it a trial.
  • Nothing is created until you confirmSubmitting the form sends a link. Your tenant and API key are created when you click it, and the key is shown once.
  • It stops, it does not billAt 10,000 calls the API returns a clear error pointing at the portal. Your key and integration stay as they are — choosing a plan resumes the same key.

Start your trial

Already have an account? Sign in.

A company address — we send the confirmation link here.

10,000 API calls across all five live agents. No card, one trial per company. We never share your details with third parties.

What counts as a call

No surprises in the billing

You should be able to predict your bill from your own logs. These are the rules the meter actually applies.

A call is one API request

One request to a documented /v1 endpoint. Submitting a lead is a call; reading it back is a call.

What is never counted

Health checks, the API reference, webhooks we send you, and every page of the customer portal.

Failures on our side are free

A 5xx from us is not billed. A 4xx caused by a malformed request is, because the work was done.

Overage is billed, not blocked

A paid plan keeps working past its allowance and bills the difference per 1,000 calls. Nothing stops mid-workflow.

The free trial stops

It has no overage to bill, so it returns a clear error at 10,000 calls rather than running up a cost you never agreed to.

Usage is visible as it happens

The customer portal shows calls against your allowance, the running total for the period and a per-endpoint breakdown.

Questions

Before you ask

Why is the trial one per company rather than one per person?

Because the allowance is real compute. We check the email domain so a trial belongs to an organisation, and send a link to confirm the address is yours. Personal and disposable mailboxes are refused for the same reason.

What happens when the trial runs out?

Calls return a clear, non-retryable error pointing at the portal, and your API key and integration stay exactly as they are. Choosing a plan resumes the same key — there is nothing to rewire.

What is a model tier?

Which model answers your calls. Essential is fast and suits well-defined work like classification and extraction; Professional reasons before answering and is the production default; Advanced is for the hardest cases. Changing tier is a plan change, not an integration change.

Do cheaper plans get fewer agents?

No. Every plan includes all five live agents. What differs is the model tier answering and the volume included.

Can we run this in our own environment?

Yes, for Scale and above. The platform is deployed per environment with its own configuration, connectors and data retention. Talk to us about what that looks like against your requirements.

How is this billed?

Monthly, in arrears, against the usage the portal shows. Nothing is charged through the portal itself — it is the record of what is owed, and your CrewTAC contact handles payment.

Next step

Not sure which plan fits?

Tell us the workflow you want to automate and the volume behind it, and we will tell you which plan covers it — or that the free trial is enough for now.

Or email sales@crewtac.com