AI integration costs depend on scope and the company’s systems. This article explains price factors, not a universal rate. It also explains how to agree a fixed amount before commissioning each phase.
Why we do not publish a single rate
Two AI assistants can have different costs. The model, system connections, data condition and usage volume vary. The impact of an error also determines the controls needed.
A rate without a defined scope does not let you compare those differences. We publish the terms we charge under and quote each phase after defining its scope.
What determines the price
- How many processes are involved. A focused use case doesn’t cost the same as several chained together, even if they share technology.
- Which systems it has to integrate with. Reading an email inbox isn’t the same as writing to the ERP, and every system has its own way of connecting.
- Data condition. We review quality and access. Scattered or incomplete data may need preparation before use.
- Error cost. Risk determines the rules, tests and validation needed. A person reviews high-impact decisions. The budget includes these controls.
- Volume and use. How many queries, documents or users: this mainly affects the operating cost.
- Security and compliance. Traceability, logs and what the EU AI Act, the GDPR and other applicable rules mean for the process.
Separate build and operating costs. Building is paid for by phase. Monthly model and infrastructure usage appears on your cloud bill. That usage cost exists even without our maintenance service. Maintenance and improvements with us are optional and quoted separately.
How we agree a price for each phase
- Free 30-minute call. We review the processes that take the most work. We then send a starting point, scope and indicative price range in writing.
- The report, before integrating AI. The AI integration report is paid work at a fixed price. It analyzes processes and data. It includes a first-phase quote and monthly operating cost. It is 100% credited against the production phase if you contract it within 6 months.
- Fixed price per phase. We agree scope, timeline and price in writing before the pilot, production and optional operation. You can stop after any phase. You keep the deliverables.
- Changes, quoted first. Anything outside the scope is quoted before it’s done. There are no open-ended hours.
Custom tools, analytics, training and the fractional AI architect do not require a report. You can order them directly, with scope and price fixed before work starts. AI integration does start with the report.
What to look at when comparing quotes
- That the scope is the same. Integration, data preparation, team training and operation: what one quote includes, another charges for separately.
- The monthly cost of running it, not just of building it. A system that’s cheap to build can be expensive to maintain.
- What happens if the scope changes: who decides and how it’s quoted.
- Who owns the code and the infrastructure. Lock-in to a provider is also a cost; it just arrives later.
- How results are measured. Agree a baseline measurement and comparison indicators before starting. We explain this in our way of working.
Before commissioning work, check the price, scope, exclusions and operating cost. Each phase must have these terms in writing.