Fixed-price sprint
An AI proof of concept in two weeks — with numbers, not a demo video.
Before you fund an AI project, find out whether it works on your data, how accurate it is and what it costs per request. One use case, two weeks, a fixed price and a clear go or no-go.
AI Proof-of-Value Sprint
Fixed scope, fixed price
AI Proof-of-Value Sprint
Find out in two weeks whether an AI use case is worth building.
Fixed price
$7,500
What's included
- Use-case scoping with measurable success criteria
- RAG or agent prototype on your own data
- Evaluation set with real questions and expected answers
- Cost model per request and per month at your expected volume
- Data handling under our AI data principles and a DPA
You receive
A go/no-go report with evaluation results, risks, the cost model and — if it's a go — a plan and estimate for production.
Who it's for
- You have a concrete use case — support answers, document search, data extraction — and need evidence before a budget decision.
- Management asks what an AI assistant would cost in operation and how reliable it is.
- A pilot with a generic chatbot disappointed, and you want to know whether your own data changes the result.
Free checklist
Scoping a RAG or AI assistant project
Use case, data sources, access control, retrieval, evaluation and costs — the questions to answer before building an AI assistant.
How the sprint runs
- 01
Use case and success criteria
Together we pick one use case and define what 'good enough' means in measurable terms.
- 02
Prototype on your data
A RAG or agent prototype on a representative sample of your documents, under our AI data principles and a DPA.
- 03
Evaluation
An evaluation set with real questions and expected answers — accuracy, failure cases and their causes, not impressions.
- 04
Go/no-go report
Results, risks, the cost per request and per month at your volume and, if it's a go, a plan and estimate for production.
Questions about the sprint
What does an AI proof of concept cost?
This sprint has a fixed price, shown above, for one use case on a sample of your data. Model API usage during the sprint is billed at cost or runs on your own provider account.
Is our data used to train AI models?
No. We use providers' commercial APIs, which do not train on API data by default, and sign a data processing agreement (Art. 28 GDPR). Details are on the AI data page.
What if the result is a no-go?
Then you have saved the cost of a full project and know why: the report names the causes — data quality, retrieval, the task itself — and what would have to change.
Which models do you work with?
Claude, OpenAI models and open-source models, chosen per use case by accuracy, cost and data-protection requirements.
Have a project in mind?
Tell us what you're building. You get an honest assessment, a clear scope and a fixed-price or milestone proposal, usually within a few working days.
Prefer to write first? Write to us