Typical ranges
| Project | Range | Duration |
|---|---|---|
| Focused automation | £1,500–£6,000 | 1–3 weeks |
| Integration between two systems | £4,000–£15,000 | 4–8 weeks |
| API or data service | £8,000–£25,000 | 6–12 weeks |
| Document processing with review | £15,000–£35,000 | 8–14 weeks |
| AI service with evaluation | £12,000–£40,000 | 8–16 weeks |
Ranges are wide because the surrounding systems vary enormously between projects that sound identical.
What moves the price
- How many systems it has to talk to, and whether they have APIs
- Data quality at the source — messy input costs real time
- How much review interface is needed
- Volume and reliability requirements
- Whether correct can be defined, for anything AI
The last one is specific to AI work. A project where nobody can say what the right answer is for a hundred cases has no way to be measured or improved.
What should be included
- Error handling and retries, not just the happy path
- Monitoring and alerting, including on absence
- Tests around the transformations and failure paths
- A deployment process
- Documentation and handover
A quote missing those is quoting a script rather than a system, and the difference appears later as unreliability.
Running and maintenance
Hosting for a typical service is £20–£150 a month. Model costs for AI work add £80–£600. Maintenance is 15–25% of build annually, higher for anything with fragile external sources.
Budget the maintenance explicitly. Unmaintained automation fails silently and the data it produces becomes wrong before anyone notices.
Start small where you can
A focused first automation at a few thousand pounds proves the approach and produces a much better estimate for anything larger.
It also builds the monitoring and deployment patterns, which makes the second project cheaper.