Pricing that fits your deployment.
We don't ship you a license and disappear. Every engagement starts with engineers from our team paired with yours — to scope what AI agents can do for your business, build the first ones with you, and hand over something already running in production.
Every tier ships with engineers from our team embedded with yours — not a CSM with a Notion page. We help you scope use cases, co-build workflows, and train your team to keep going on their own.
Most teams don't need another agent framework. They need to figure out where AI fits in their actual business. We bring opinions: what to automate, what to leave alone, where governance has to sit.
Models billed at provider list — pass-through, no markup. Reviewers and approvers don't count as seats; you'd be punished for governance otherwise. Engineering and admin seats only.
Engineers paired with yours, start to finish.
Buying a control plane doesn't make you good at AI. The hard part is figuring out which workflows are worth automating, what your policy team will accept, and which model fits which job. We sit alongside your team and work through it together.
Scope use cases that pay back
Workshops with your engineering, ops, and compliance teams. We map out a short list of agent candidates, rank them by impact and feasibility, and pick the first to ship.
Build the first agent together
A senior engineer from our team writes the workflow alongside yours. Policies, evals, audit, approvals — all wired in. By the end, your team can repeat it without us.
Quarterly strategy reviews
Once you're running, we review every quarter: what's working, what's drifting, what new use cases to take on. Strategic, not status-update.
Real engineers
Not CSMs. Not contractors. The engineer in your Slack also writes our product code.
We say no
If a proposed use case won't work, we tell you. We've watched teams burn a year on the wrong agent.
Patterns from 40+ rollouts
Claims triage, KYC, contract review, clinical summarization, IT triage — we've shipped most of the obvious ones.
Your team owns it
By the time we step back, your engineers can ship the next workflow themselves. That's the whole point.
Three engagement shapes.
Most teams start in Pilot, move to Operate once they have a regulator-approved workflow, then graduate to Founding Customer when the control plane spans business units. We don't gate compliance behind tiers — every tier ships SOC 2, hash-chained audit, residency.
Pilot
A named solutions engineer and a 2-week discovery sprint to identify the right first agent. We build it with you, against your data, and stand it up in staging. Convert to Operate or walk away.
- 3 agents and 1 workflow in staging
- SaaS-hosted in your region
- Frontier model routing (Anthropic, OpenAI, Google)
- Hash-chained audit with 14-day retention
- SOC 2 Type II covered, BAA on request
- Named solutions engineer for the duration
- 2-week procurement white-glove (NDA, security questionnaire, draft MSA)
- Single-tenant VPC deployment
Operate
You're past pilot. The agent is reviewed, the policies are written, you need it to run reliably under your SOX / HIPAA / FINRA posture.
- Unlimited agents and workflows
- SaaS or single-tenant VPC (AWS, GCP, Azure)
- Any frontier model, BYO endpoints, cost budgets
- Hash-chained audit with 7-year WORM retention
- HIPAA BAA, GDPR DPA, EU AI Act mapping
- Migration credit off LangChain or homegrown stacks
- Source-code escrow for your procurement team
- Shared Slack channel, 4-hour SLA, quarterly architecture review
Founding Customer
Our first 25 enterprise customers. You shape the product; we put your name on it. For multi-BU rollouts, government, and anything needing redlined MSA before the first call.
- Everything in Operate, plus:
- On-prem and air-gapped deployments
- Open-weights and BYO model weights
- Named TAM, executive sponsor, quarterly business reviews
- 1 hour per quarter with our CEO or CTO, guaranteed
- Named-date commitments on your roadmap items
- Co-marketed launch and reference customer status
- Procurement-grade paper: MSA, DPA, BAA, sub-processor map
Don't see your shape? We've also written one-off engagements with research orgs, public-sector pilots, and design partners. Tell us what you'd want a relationship to look like — we'll tell you whether it fits.
What's actually different.
Compliance is the floor everywhere. The differences are about deployment, scale, and the shape of our relationship.
| Pilot | Operate | Founding Customer | |
|---|---|---|---|
| Workloads | |||
| Agents in productionNumber of governed agents allowed to serve real traffic. | Up to 3 · staging | Unlimited | Unlimited |
| WorkflowsMulti-step orchestration with gates, branches, write-backs. | 1 | Unlimited | Unlimited |
| Runs per monthSoft cap; hard cap negotiated on contract. | 25k | Per quote | Per quote |
| Deployment | |||
| SaaS · multi-tenantus-east-1, eu-west-1, ap-southeast-1. | ✓ | ✓ | ✓ |
| Single-tenant VPCYour cloud, your IAM, your KMS — AWS / GCP / Azure / Oracle. | — | ✓ | ✓ |
| On-prem · air-gappedOpen-weights models, signed install bundle, sovereign. | — | — | ✓ |
| Compliance | |||
| SOC 2 Type II report (NDA) | ✓ | ✓ | ✓ |
| HIPAA BAA · GDPR DPA | On request | ✓ | ✓ |
| Hash-chained audit retention | 14 days | Up to 7 years | Up to 7 years · WORM |
| Custom policy DSLType-checked rules, dry-run, versioned alongside your code. | — | Beta | ✓ |
| EU AI Act risk-tier mapping | ✓ | ✓ | ✓ |
| Support & relationship | |||
| Solutions architect paired | ✓ | ✓ | ✓ |
| Response SLA · email | 1 business day | 4 business hours | 1 hour · 24/7 paging |
| Shared Slack / Teams channel | — | ✓ | ✓ |
| Named TAM & QBRs | — | — | ✓ |
| Account-level roadmap commitments | — | Influence | Named dates |
| Commercial | |||
| PaperWhose templates the contract is built on. | Kommit standard | Either | Your MSA |
| Term | Up to 60 days | Quarterly · annual | 1–3 years |
| Model spend | Pass-through | Pass-through | Pass-through or pooled |
Optional, quoted on request.
A few things customers ask for that we keep priced separately so they don't get baked into a base they don't need.
White-glove rollout
A senior engineer embeds for 2–8 weeks. We build your first three workflows alongside your team, then hand over with a runbook.
Auditor support
We sit alongside your auditor (Big Four or internal) and answer their questions for SOX, SR 11-7, EU AI Act, or your FINRA / OCC exam.
Workshops for your team
Half-day or full-day deep-dives for engineering, compliance, or operations — taught by the engineers who built the relevant piece.
Connectors we don't ship
If your stack of record is unusual, we'll quote a connector and ship it under the same SLA as our native ones.
Sovereign deployments
Signed install bundle, offline catalog, on-site rotation of an engineer for the initial install. Common for government & defense.
Eval suite as a service
Our eval team helps you build the golden suite for your most sensitive workflow. Domain-aware, attestation-ready.
How to get a number, fast.
From first call to signed quote, typically two weeks. We don't drag it out.
- Discover
30-minute briefing
Solutions architect + engineer joins. We ask about your stack, your compliance posture, your timeline. No deck on our side.
no charge - Scope
Written proposal · 3 business days
Deployment shape, workflow estimate, support tier, contract term. Plain English, single page.
gate · we'll say no if we don't fit - Paper
NDA · DPA · MSA
Your paper or ours. Standard NDA returned within 24h. Security questionnaire returned within 4 business hours.
gate · counsel signoff - Pilot
Live workflow in staging
14 days, your data, against your real systems of record. Convert to Operate or walk.
commits begin here
What we get asked first.
Why no public prices?+
Because we'd be lying if we put one number on a page. A pilot for a single workflow at a small fintech and a multi-business-unit rollout at a top-10 carrier aren't comparable. We'll get you a written quote within three business days of a discovery call — faster than most published pricing pages take to find the right tier anyway.
How do you charge for model usage?+
Pass-through. The router calls the provider with your account (or our negotiated account if you use ours), and the provider's invoice flows directly to you or via a single line item from us. Zero markup. We make our money on the platform fee, not on tokens — which is also why our incentives line up with you using cheaper models when they pass eval.
Do reviewers and approvers count as seats?+
No. Engineering and admin seats only. We'd be punishing customers for governance otherwise — and dual-control approvals would be the most expensive operation in the product. The whole point is to make oversight cheap.
Is the pilot binding?+
No. Pilot is a fixed fee with a defined scope, and at the end you decide whether to convert. About 98% of pilots do, but the 2% who don't walk away cleanly. The pilot fee is non-refundable but always credited toward the first quarter of Operate if you continue.
Can we start with Founding Customer paper from day one?+
Yes. If your procurement requires an MSA before any work begins, we'll negotiate the MSA in parallel with the pilot. Allow 1–3 weeks for legal turnaround, depending on your redlines. Standard MSA fits on 8 pages.
What's the multi-year discount?+
Modest — 5–10% — and we'd rather you sign a 1-year term and renew willingly. We don't do steep multi-year discounts because we don't want to lock in a price before we know what your usage will look like.
Do you have a non-profit or research discount?+
Yes. For accredited universities, research labs, and 501(c)(3) organizations doing AI governance work, we offer a 50% discount on Operate and waive the pilot fee. Tell us about your work in the first call.
Tell us your scope.
30 minutes with our team gets you a written quote within three business days — deployment model, workflow estimate, support tier, contract term. No drip sequence, no SDR call.