Pay As You Go AI Assistant: Only Pay When It Does Something

The pay as you go AI assistant is a specific pricing model — and it is the right model for a specific kind of user. Understanding when it makes sense and when it does not is more useful than blanket claims about which model is cheaper.

Flat subscription pricing exists for a reason: if you use a tool every day and extract consistent value, a predictable monthly cost is fair and often efficient. The math breaks down when usage is variable. If you pay $20/month for an AI assistant but only actively use it twelve days out of thirty, you are paying $1.67 per day of actual use — not the advertised $20/month. Pay-as-you-go pricing means the cost follows the value, not a calendar.

Definition — Pay As You Go AI

Pay as you go AI pricing is a billing model where you are charged per action taken or per unit of AI usage, rather than a fixed monthly fee regardless of how much you use the product. Under this model, you pay when the AI does something — sends an email, books a meeting, runs a research query — and pay nothing for idle time. It is the same model used by cloud infrastructure (AWS, Google Cloud) applied to AI assistant services.

How Deputy's pricing works

Deputy is built on ATXP — an agent infrastructure platform that provides identity, payments, and tool access for autonomous AI agents. ATXP's per-action billing model flows through to how Deputy charges users.

When Deputy takes an action — sends a follow-up email, schedules a calendar event, runs a research task, fires a reminder sequence — that action incurs a cost. When Deputy is idle between tasks, there is no charge. Signing up is free and there is no monthly minimum.

The "$1/day" framing is the typical usage anchor for a user running a consistent daily set of tasks. If you use Deputy lightly, you will spend less. If you route significant workflow volume through it, you will spend more — but you are also getting more done. The cost tracks the value, which is the point.

When pay-as-you-go pricing wins

Pay-as-you-go pricing has a clear advantage in three specific situations.

Light users. If you want an AI assistant available but do not need it every single day, a flat $20/month subscription means you pay for days you do not use it. Pay-as-you-go pricing makes the always-available AI economically rational for lighter usage patterns.

Variable usage. Some weeks are heavy — lots of follow-ups, scheduling complexity, research tasks. Other weeks are quiet. A flat subscription does not flex with your usage. Pay-as-you-go does — heavy weeks cost more, light weeks cost less, and the total tracks what you actually needed.

Multiple subscriptions already. If you are already paying for ChatGPT, Claude, or other tools, adding another flat $20/month subscription is a real cost increase. Adding a pay-as-you-go tool that costs less than $1/day when used is a different calculation — especially if Deputy handles the action-taking tasks those other tools cannot.

When a subscription makes more sense

Subscription pricing wins for heavy daily users with consistent, predictable use cases. If you open ChatGPT every single workday, run multiple sessions per day, and the tool is genuinely embedded in your daily workflow, $20/month is efficient — your per-use cost approaches $0.06/session.

Subscription also makes sense when you want completely predictable billing. Some users and businesses need to know exactly what they will spend each month for budgeting purposes. Variable pay-as-you-go pricing, even when it averages out lower, introduces uncertainty that some people prefer to avoid.

Cost comparison: flat subscription vs. Deputy's less-than-$1/day pricing

The table below compares the three most common AI subscription options against Deputy's usage-based pricing for typical users.

The math above does not even account for the full subscription stack. According to the Productiv 2025 SaaS Benchmark Report, the average knowledge worker pays for more than three AI subscriptions simultaneously — which means the real annual cost of an AI stack is often $600–$900/year before any enterprise tools. For more context on this, see our full breakdown of the real cost of AI subscriptions.

"Anything that the user can't do themselves. Don't sell the future."

The honest trade-off between the two models

Pay-as-you-go pricing is not strictly better than subscriptions. It is the right model for the right user. The honest evaluation looks like this.

If your usage is heavy and daily — you are getting real work done with the tool every single day — a flat subscription is likely more economical because your effective per-use cost drops as usage increases. ChatGPT Plus at $20/month across 200 daily uses is $0.10 per use. That is hard to beat with usage-based pricing at any meaningful usage level.

If your usage is variable — you need an AI assistant available but do not engage with it every day — pay-as-you-go pricing means your costs track your actual usage. Three light weeks and one heavy week will cost you less than four weeks of flat subscription. For users who have the relevant tools but are not fully utilizing them, see our overview of the best AI assistants with no subscription.

There is also a category difference worth noting. Most flat-subscription AI tools (ChatGPT, Claude) are primarily generative — they produce content you then act on. Deputy is action-taking — it executes tasks. You are not just paying for a different pricing model; you are paying for a fundamentally different type of work being done. For a broader comparison of the tools available, see the best AI assistants in 2026.

Frequently asked questions