UTENX · ENGINEERING
AI automation cost: what changes the scope and budget?
Price the connected workflow, not just the model call. Systems, access, exception handling and operating ownership determine the build.
Start with the outcome
AI automation cost depends on what must happen between an input and an accepted business result. A model that produces an answer is one component. Reading customer context, getting permission, updating another tool and making failures visible are part of the workflow.
Begin with a specific task, its owner and the evidence that the task has completed correctly. A narrow, well-defined process is easier to scope than a request to automate an entire department. This guide explains the cost drivers Utenx uses to discuss a project; it is not a market-wide price survey.
The current Utenx engagement model
| Engagement | Published price | Purpose |
|---|---|---|
| Map | $8k | One week: inventory, gaps and a build scope |
| Connect | $45k–$120k | Fixed-scope build, generally six to ten weeks |
| Optional retainer | $12k/month | Reserved capacity as the stack changes |
The proposal confirms currency, terms, third-party expenses and the agreed delivery date. The retainer is optional. These are Utenx engagement prices, not average prices for the market or a quote for a particular automation.
Five scope drivers
- Connections: how many systems are involved, and which operations their interfaces support.
- Data readiness: how customer records, identifiers and missing fields are resolved.
- Access: who approves credentials and which permissions the workflow requires.
- Exceptions: what happens when inputs are unclear, a tool fails or a person must review a decision.
- Acceptance and handoff: how the team verifies delivery and operates the result.
Two workflows with the same model can have very different integration costs. A read-only assistant with existing data access is different from a workflow that changes financial or customer records across several systems.
Include running costs
List the third-party products, model usage, hosting and monitoring that the design needs. Use expected workload and task behavior to form assumptions; do not extrapolate from an unrelated demo. Keep recurring product fees distinct from engineering delivery and any optional support engagement.
An illustrative workload estimate should state the number of tasks, typical input size, retries and human review. Keep these as editable assumptions until observed usage replaces them. No fixed savings percentage or payback period is implied by a build quote.
Compare value with a disclosed baseline
Measure the workflow before changing it: work volume, time spent, exceptions and rework. Define the comparison period and account for work that remains with a person. More completed tasks are not valuable if the result needs costly correction.
Treat any ROI worksheet as a decision aid. The team can compare delivery and running costs with a real, documented operational baseline after a pilot or go-live. The indicative AI readiness tool can help frame the starting questions, but it is not a quotation.
What to bring for a useful estimate
- The task and the business outcome you need.
- Systems involved and the person responsible for access.
- Representative inputs, including unusual cases.
- Work volume and current handling time, if measured.
- Acceptance requirements, budget context and deadline.
A paid Map engagement can turn an uncertain stack into a specific build scope. Explore AI automation services and the existing engagement options.
Related work and guidance
NEXT STEP
Scope the system you need.
Bring the workflow, tools and result your team needs. Start with a 30-minute project scoping call or send a short brief.