Separate the initial work from the running cost
A quote for “an AI assistant” is hard to compare unless you know what it includes. The same description could mean a reusable prompt, a draft-generation workflow, or software that reads and updates several business systems. Those are different projects.
For AlexGPT, a one-hour consultation is $59 USD and system design is $299 USD. Implementation is scoped separately. Those are service prices, not a claim about the typical market price or the total cost of your project.
Ask for a budget with these lines
| Budget item | What to clarify |
|---|---|
| Discovery and design | Which workflow, deliverables, and decisions are included? |
| Implementation | Which systems connect, and what is excluded? |
| Software and model usage | Fixed subscription, usage-based charges, or both? |
| Human review | Who checks results and handles exceptions? |
| Maintenance | Who fixes failures when tools or processes change? |
| Handover | Does your team receive instructions and access? |
Calculate net capacity, not imaginary revenue
Start with: monthly value of time released − monthly software, review, and maintenance costs. For a one-time setup cost, divide it by a positive monthly net benefit to estimate a simple payback period. If the benefit is zero or negative, there is no payback under those assumptions.
Illustrative example: a workflow releases 20 hours per month valued internally at $20 per hour. That represents $400 of capacity. If ongoing costs total $150, the estimated monthly benefit is $250. A hypothetical $1,000 setup would take four months to recover. These numbers are invented for the calculation; they are not a quote or a client result.
Time released is not automatically cash saved. If salaries stay the same, the team needs valuable work to do with the available capacity. Also consider whether faster responses, fewer errors, or better records matter enough to justify the work.
What makes the estimate less predictable?
Messy source data, missing integrations, unclear rules, variable document formats, and expensive mistakes all increase uncertainty. A pilot helps reveal the cost of exceptions before a larger commitment. Ask what happens when the tool is unavailable or a required field is missing.
When to start smaller
If you cannot describe the workflow or measure the baseline, begin with a consultation. If the process is clear but the architecture is not, a design can define the scope. If an existing feature already solves the problem, enabling it may be more sensible than commissioning a build.
One process. A useful next step.
Bring the workflow you want to improve. We’ll discuss where AI can help, what needs checking, and what to do next.
Book a consultation — $59 ↗