Investment

The only benchmark that matters is the one you run yourself.

Because the most expensive and risky answers are the ones that sounded right.

MID MARKET PILOT

$7500/  4 weeks 

Founding community pricing. For an organisation with a defined use case and an engineering team ready to evaluate. 

N

Onboarding call to evaluate design and success metrics agreed in writing ie: target token savings, edge use case scenarios

N

4 deployments (dev, staging, pre-prod, prod)

N

Direct Open AI compatible API access: Claude Code, VS Code, Cursor, or any stack

N

Weekly check-in during the pilot period via Slack

N

End-of-pilot results review meeting with Comfrey team

N

Universal Agentic Reasoning Layer API access

N

Deployment scripts for any cloud / open model

N

Slack support

N

Pilot fee credited toward annual license

N

Money back guarantee should your success metrics not be met during the pilot

TALK TO US

enterprise pilot

$15000/ 6 weeks

Founding community pricing. For regulated industries and complex deployments that need certainty before committing. Founder-led engagement throughout.

N

90-minute kickoff to evaluate design, success metrics agreed in writing, integration planning

N

4 deployments (dev, staging, pre-prod, prod)

N

Direct Open AI compatible API access: Claude Code, VS Code, Cursor, or any stack

N

Weekly 60 minute meetings with Comfrey CTO throughout the 6 weeks

N

Direct Open AI compatible API access: Claude Code, VS Code, Cursor, or any stack

N

Custom deployment support

N

Universal Agentic Reasoning Layer API access

N

Deployment scripts for any cloud / open AI model

N

Success metrics agreed on day one

N

90 minute kick off meeting

N

Weekly meetings for 6 weeks (30 minutes)

N

Named contact

N

Security review support

N

Unlimited trouble shooting

N

CAD invoicing available, upon request

N

Money back guarantee should your success metrics not be met during the pilot

TALK TO US

These are our founding prices.

We are opening UARL to a limited founding community, the organisations and teams willing test our layer before it becomes the obvious choice. When that community is complete, these prices will no longer be available.

We have always believed that the people who choose you early deserve something more than a receipt. We look after the people who decided to join what we are building and said yes before the rest of the world caught up.

Have a Question?

Contact US
Will UARL work with the AI model we are already running: Claude, ChatGPT, Gemini, Cohere?

Yes. UARL is model-agnostic by design. It needs access to the weights so it works effortlessly with open models as a reasoning layer on top of whichever model or stack you are already using you do not need to replace, retrain, or reconfigure your existing AI stack. During the pilot, you will connect UARL to your current model via an OpenAI-compatible API, which covers the models most enterprise teams are running today.

Do we need to replace our existing RAG system to run the pilot?

No. UARL sits on top of your existing RAG system, it does not replace it. During the pilot, you will have API access to connect UARL between your current retrieval layer and your model, adding the reasoning loop that standard RAG is missing. Your documents, your vector store, your retrieval configuration, all of it stays in place. UARL adds the step between retrieval and generation that tests every claim before it reaches your model. If your RAG system is already working well, UARL makes it more accurate. If it is struggling with hallucinations or inconsistent answers, your compute bill is rising then UARL is the architectural fix.

Does the pilot require us to send any data to Comfrey's servers?
No. UARL deploys entirely within your virtual private cloud. Your documents, prompts, queries, and outputs never leave your infrastructure during the pilot or at any point after it. The only outbound signal is a lightweight license check that runs on startup and every 24 hours it sends two things only: model ID and license expiry status. No prompts. No completions. No document content. No user data. Comfrey never sees what your AI is doing with your data. That is not a policy commitment. It is an architectural one.
What happens at the end of the pilot?

Three outcomes: if you hit the success metrics we agreed on day one, the pilot fee is credited toward your annual contract. If you need more time, we can extend at $2,500/month. If it’s not the right fit, the engagement ends, no obligation. The pilot is available once per organisation.

What exactly happens during the pilot, week by week?
Week 1: Onboarding call where Comfrey works with your team to define success metrics in writing such as target hallucination reduction, token savings, specific edge case scenarios you need UARL to handle. UARL is deployed across your dev and staging environments.

Weeks 2–3: UARL runs against your defined use case. Your engineering team integrates via the OpenAI-compatible API. Comfrey is available on Slack throughout. Weekly check-in to review early results and address any integration questions.

Week 4 (mid-market) / Weeks 4–6 (enterprise): Full deployment across pre-prod and production environments. End-of-pilot results review with the Comfrey team — measured against the success metrics agreed on day one. If the metrics are met, the pilot fee is credited toward your annual license should you wish to continue. If they are not, the money-back guarantee applies.

What does the money-back guarantee actually cover?
If UARL does not meet the success metrics agreed in writing on day one of the pilot, you get the pilot fee back. No partial credits, no extended engagement to try again, no negotiation. The metrics are agreed before any work begins, which is why defining them carefully at the outset matters. This guarantee exists because Comfrey’s founding position is that the only benchmark that matters is the one you run yourself. If UARL cannot prove its value on your data, with your model, against your use case, you should not pay for it.
Is CAD invoicing available?

Yes, for Canadian public sector and government procurement. All prices on this page are in USD by default. If your organisation requires CAD-denominated contracts or invoicing, let us know when you book a call and we’ll handle it from there.