AI investment
What an AI automation project costs
Our planning prices by scope, what each band includes, and the evidence we need before giving you a fixed quote. Updated for 2026.
TL;DR
- Some Tech Work charges €3,500 to €5,000 plus VAT for fixed-price workflow discovery. You receive a process map, risk register, acceptance criteria, and a fixed build proposal.
- A single production workflow usually falls in our €8,000 to €20,000 planning band when it has one primary system, stable access, and a bounded exception path.
- A model-assisted or multi-system workflow usually falls in our €20,000 to €40,000 band. More systems, permissions, input variation, and approval risk move the quote upward.
- Multi-workflow programmes are phased after discovery. We do not publish a universal enterprise price because the scope is not comparable.
- These are Some Tech Work planning bands. They do not represent a market average or a quote. Third-party software, model usage, tax, and ongoing support are separate unless the proposal includes them.
What you are actually paying for
A model call is one line item. The operating system around it is the project.
A demo proves that one happy path can produce output. A production workflow needs permissions, input validation, error handling, retries, alerts, a review queue, a manual fallback, deployment, and an accountable owner. Those requirements determine the scope.
The integration and operating boundary usually drives the cost. We price the systems touched, the variation in real inputs, the actions the workflow may take, and the evidence needed for acceptance. Model access is only one input.
Tool selection follows those constraints. A configured workflow between stable SaaS APIs can be smaller than a custom service handling legacy systems or retrieval over controlled documents. Ask any vendor to name the inputs, writes, exceptions, acceptance test, handover, and ongoing owner in the proposal.
What makes a production AI system expensive is not the AI model. It is the integration surface, the data preparation, the error handling, the monitoring, and the documentation that makes it auditable and maintainable.
Cost by project scope
Some Tech Work planning bands for AI automation.
| Engagement | Our planning band | Included | Quote assumptions | Output |
|---|---|---|---|---|
| Workflow discovery | €3,500 to €5,000 | Process map, data and access review, risk register, acceptance criteria, delivery plan | One candidate workflow and access to the people who run it | Written scope and fixed-price build proposal |
| Single production workflow | €8,000 to €20,000 | Build, test, deployment, monitoring, exception route, runbook, handover | One primary system, stable access, bounded inputs and approvals | One accepted workflow running in production |
| Model-assisted or multi-system workflow | €20,000 to €40,000 | Evaluation set, model or retrieval layer, integrations, human review, monitoring, runbook, training | Defined data boundary, test cases, system owners, approval and fallback rules | Production workflow with measured acceptance results |
| Multi-workflow programme | Phased quote after discovery | Roadmap, dependency order, shared controls, programme reporting, phased delivery | Accountable executive owner, funding gates, system access, adoption plan | A sequence of separately accepted production releases |
At a glance
- 3,500–5,000€
- Our fixed-price discovery band before a build commitment.
- 8,000–20,000€
- Our planning band for one bounded production workflow.
- 20,000–40,000€
- Our planning band for a model-assisted or multi-system workflow.
What drives the price
Four facts that move a workflow between planning bands.
Data quality and preparation
We sample real inputs before quoting. Consistent fields and known formats reduce mapping and test work. Scans, missing identifiers, conflicting records, and undocumented labels add a preparation workstream that appears separately in the scope.
Custom LLM work versus configured tools
We use configured workflow tools when they meet the access, error-handling, data-location, and ownership requirements. We use a custom service when the workflow needs controls or behaviour the platform cannot provide. The quote states the choice and its maintenance consequence.
Human review and audit requirements
Every approval point needs a queue, permissions, context for the reviewer, an override path, and a record of the decision. The right boundary depends on the action and applicable rules. Privacy, legal, security, or works-council review stays with the accountable specialist on the buyer side.
Number of system integrations
Each system brings credentials, permissions, API limits, test environments, ownership, and failure modes. Complexity does not rise in a neat line: one legacy write path can cost more to validate than several stable read-only APIs. We inspect the interfaces before fixing the price.
Automation next step
Scope this as an automation: AI automation project cost
Tell us the manual step you want gone. We reply with what is realistic to automate and what it costs.
Common questions
What decision-makers ask when budgeting an AI automation project.
How much does an AI automation project cost?
At Some Tech Work, discovery is €3,500 to €5,000 plus VAT. One bounded production workflow is planned at €8,000 to €20,000. A model-assisted or multi-system workflow is planned at €20,000 to €40,000. These planning bands do not represent quotes or market averages. A written proposal fixes the number, assumptions, inclusions, third-party costs, and acceptance criteria.
What is the difference between an AI pilot and a production AI system?
A pilot tests a hypothesis on bounded data. A production system has controlled access, validation, failure handling, alerts, monitoring, a manual fallback, ownership, and an acceptance record. We quote those requirements directly rather than applying a universal pilot-to-production multiplier.
Do AI automation projects require ongoing costs after delivery?
Usually. Third-party licences, model or cloud usage, monitoring, incident response, evaluation, and changes to connected systems can continue after launch. The proposal separates estimated external consumption from an optional support scope. Ask for the unit, expected volume, alert threshold, and owner of every recurring cost.
What workflows are best suited for AI automation?
Workflows with high volume, repetitive structure, and clear acceptance criteria automate well. Examples: invoice data extraction, lead qualification routing, support ticket classification, contract clause review, report generation from structured data. Workflows that require nuanced human judgment, manage sensitive relationships, or have low volume and high variability are poor candidates. Start with repetitive, high-volume workflows where acceptance criteria are clear.
Does AI automation comply with GDPR?
No generic architecture proves compliance. Document the purpose, data sources, affected people, vendors, storage and transfer locations, retention, automated actions, approval and override paths. Then have the accountable privacy, legal, security, or works-council owner review the specific use case. Our scope records technical controls and evidence; it does not replace legal advice.
How we scope AI projects at Some Tech Work
We scope AI automation in a fixed-price discovery before any build commitment.
Our AI automation projects start with fixed-price discovery: process mapping, input and access sampling, risk and exception mapping, acceptance criteria, and a written scope. Discovery costs €3,500 to €5,000 plus VAT and produces a fixed-price build proposal with explicit assumptions and deliverables. The schedule is confirmed in that proposal because access and test data determine it.
We build on open-source automation platforms for standard workflows and custom services for LLM-intensive or retrieval projects. All AI projects include production monitoring, a human review path, and handover documentation. See our AI automation delivery capability for scope and case examples.
Opportunity scorecard
Score the workflow before anyone approves an AI pilot.
Free resource
Download the AI automation opportunity scorecard
Score each workflow across eight criteria: volume, acceptance rules, system access, data sensitivity, ownership, and more. Includes production guardrails template.
Loading form…
Written by
Vineet Talwar
Co-founder, Tech & Operations at Some Tech Work. WordCamp speaker across Europe and Asia, and host of the WP Shoutout podcast.
Start here
Ready to talk.Book a short diagnostic.
Tell us what needs fixing
We read every brief and reply within one business day.
Prefer to talk first?or request a tech stack audit →or email us directly →
Not sure where to start? Send the stuck decision, workflow, or page. We will say whether you need a diagnostic call, a tech stack audit, or a different first step.