Pricing and quotation guide
AI automation cost
AI automation has three kinds of cost: the one-off work to design, build and test the workflow; the running costs of the AI provider and automation platform, which usually scale with volume; and the people time to review results and maintain the workflow.
This page explains each kind of cost and how to estimate the total before committing.
Illustrative diagram
AI automation cost
- Build
- Discovery
- Workflow and testing
- Running
- AI usage
- Platform fees
- People
- Review
- Monitoring
- Decision
- Prototype first
- Pilot results
What drives AI automation cost
- Discovery
- Mapping the process, volumes and exceptions before building.
- Workflow complexity
- Number of systems connected, steps, rules and branches.
- AI step design
- Prompts, validation rules and handling of uncertain results.
- Testing
- Running real examples, measuring errors and adjusting.
- Pilot
- Running alongside the current process and reviewing results.
- Documentation and handover
- Configuration, prompts and review routine documented.
Costs paid to third parties (separate from our fees)
- AI provider usage, usually charged by volume of text processed
- Automation platform subscription or usage
- Any new software the workflow needs
AI and platform charges are paid to their providers, in your accounts, and estimated separately from build work.
Estimating running costs
| Input | How it is estimated |
|---|---|
| Items per month | From your current volumes |
| Size of each item | Measured on sample documents or messages during testing |
| AI usage per item | Measured during the prototype |
| Platform runs per item | Counted from the workflow design |
| Review time | Share of items needing a person, multiplied by time per review |
Count the review time
AI workflows usually still need people: reviewing uncertain results, correcting errors and checking samples. That time is a real cost and should be estimated honestly. A workflow that handles most items automatically but needs careful review of the rest can still be very worthwhile; one that needs a person to check every item may save little.
The prototype measures what share of items need review, which makes this estimate concrete.
Custom workflow or off-the-shelf tool
Off-the-shelf AI tool
Suits you when
- A product exists for your exact task
- Your process matches its assumptions
- You want vendor support
Watch out for
- Subscription per user or volume
- Limited control over data and logic
Custom workflow
Suits you when
- Your process or systems are specific
- You need control over data and review
- Several systems must connect
Watch out for
- Build and maintenance cost
- Someone must own it
For an accurate quotation
Share
- Description of the current process
- Monthly volumes
- Anonymised examples, including awkward ones
Tell us
- Systems involved
- Who will review results
- Any data that must not leave your systems
Example: sorting incoming enquiries
Illustrative example — not a client case study
Situation: A business receives a steady stream of email enquiries that a staff member sorts and forwards by hand.
- A prototype classifies a sample of past enquiries and measures how many it sorts confidently.
- Confident results are routed automatically; uncertain ones go to a person.
- Running cost is estimated from measured AI usage per email and monthly volume.
What this shows: The quotation set out build cost, estimated monthly usage and expected review time. This example is illustrative.
Phasing the investment
A sensible sequence is discovery, then a small prototype, then a pilot, then rollout. Each stage is quoted separately and ends with a decision point. If the prototype shows the AI step is unreliable for your data, you stop having spent little. If the pilot shows review takes longer than expected, you can adjust scope before committing to a full rollout.
This staged approach suits AI particularly well, because results depend so much on the specific data involved.
Maintenance costs
AI workflows need occasional updates when connected systems change, AI providers update their models or new kinds of input appear. Budgeting a small amount for periodic review and adjustment avoids surprises and keeps accuracy from drifting.
Comparing with the current cost
To judge the investment, compare the total of build, running and review costs over a realistic period with the cost of the time currently spent on the task, including the cost of errors and delays. Some tasks show a clear case; others turn out to be better left manual or handled by simpler automation. Knowing that before a full build is itself valuable.
Questions about AI automation cost
How is AI usage charged?
Usually by the volume of text processed, paid to the AI provider in your account.
Can you estimate monthly running costs?
Yes, from your volumes and usage measured during the prototype.
Is review time really a cost?
Yes. It should be counted when judging whether automation is worthwhile.
Is a prototype charged?
A prototype is usually quoted as a separate, smaller item so you can decide before a full build.
Are platform subscriptions included?
No. They are paid to the platform provider and estimated separately.
Will costs change over time?
Provider pricing and your volumes can change. Monitoring usage keeps surprises small.
Can we stop after the prototype?
Yes. Each stage ends with a decision, so you only continue if the results justify it.
What if AI is not worth it for our task?
We will say so. Sometimes simpler automation or a process change is the better answer.
Request a Custom Quote
Ask about AI automation cost
Thank you — your enquiry has been sent.
We have received your message and will reply by email. If you need to add anything, email info@letgetgrow.com.