AI Runs Explained: How Many an AI Form Builder Actually Uses

Short answer: an AI run is one build-time generation request — creating a form from a description, extracting fields from a PDF, rewriting a question set. Your customers' submissions do not consume them. As of September 2026 BotForms includes 100 runs a month on Starter at $29, 1,000 on Pro at $79 and 5,000 on Business at $199, and for most businesses the allowance is not the constraint that decides the plan.

Every AI-powered tool meters its model usage somehow, and every pricing page names it differently — runs, credits, generations, tokens. The unit is genuinely hard to reason about before you have used the product, which makes it hard to judge whether a plan is generous or stingy.

The good news is that the honest answer for most businesses is "you will not come close", and the more useful exercise is understanding when that stops being true.

What actually consumes a run

The distinction that clears up most of the confusion: AI work happens while you build, not while people submit.

Consumes a run:

• Generating a form from a written description
• Regenerating or refining that form after you change the description
• Extracting the fields from an uploaded PDF or scanned document
• Asking the tool to rewrite or improve a set of questions

Does not consume a run:

• A customer opening and completing your form
• Receiving a submission, or a notification about it
• Generating a document from a submission
• Exporting, filtering, or sending submissions onward

That second list is where all your volume is. A form that takes 800 submissions in a month cost you the handful of runs it took to build, once.

What a real month looks like

Building a form is bursty. You describe it, look at the result, change the description, look again — five to fifteen runs across an afternoon, sometimes more if you are converting a complicated paper form. Then the form sits there working and consumes nothing.

Rough shapes, by kind of business:

Business	Forms built or reworked per month	Typical runs used	Comfortable on
Solo trader, one enquiry form	0–1	Under 15	Starter (100/mo)
Small service business, a few forms	1–3	20–60	Starter (100/mo)
Marketing team, campaign landing pages	4–8	80–200	Pro (1,000/mo)
Agency building for clients	15–40	300–900	Pro or Business
Digitising a paper archive	Project burst	300+ in one month	Business, or an add-on top-up

The pattern in that table is the useful part. Run consumption tracks how often you build, and building is a project activity with a long tail of nothing. The month you set everything up is your peak, and it is usually your peak by a wide margin.

The exception: PDF conversion at volume

One workload genuinely eats the allowance, and it is worth flagging because it is also one of the most valuable things these tools do.

Turning a stack of paper or PDF forms into online forms is model-heavy. A long document may take more than one pass to extract cleanly, and you will iterate on the ones with awkward layouts — tables, multi-column sections, handwritten annotations. Budget several runs per document rather than one, and if you are digitising forty legacy forms in a fortnight, plan for that as a burst rather than assuming a normal month. The mechanics of that conversion, and what it does and does not preserve, are covered in converting a PDF to an online form.

The sensible approach for a one-off migration is a top-up rather than a permanent plan change — on BotForms, extra runs are $8 per 500. Paying a hundred and change more per month forever for one busy fortnight is the wrong trade.

Why the run allowance rarely picks the plan

Look at what actually differs between tiers on any AI form platform and you will find the run allowance is the least binding line on the list. On BotForms, moving from Starter to Pro changes:

Submissions from 1,000 to 10,000 a month. Storage from 5 GB to 25 GB. Users from 1 to 3. It adds team collectors, API access, unlimited PDF form tracking, email campaigns and a jump in audience size from 500 to 5,000 contacts. And it raises runs from 100 to 1,000.

For nearly every business, one of the first three numbers hits its ceiling long before the last one does. Submission volume is the usual binding constraint; storage becomes one quickly if your forms accept photos, as file and photo uploads on forms sets out in arithmetic; and seat count is what forces the move for a growing team.

So the practical advice is to size on submissions, storage and seats, then check the run allowance is not absurd for how much building you do. It almost never is. Current numbers for every tier are on pricing, and if the entry tier is what you are weighing, the cheapest AI form builder options for a small business compares that end of the market directly.

Using fewer runs without trying

Three habits cut consumption substantially, and each of them also produces a better form.

Describe the form properly the first time. The single biggest source of wasted runs is a two-line prompt producing a generic form, which then needs four regenerations to get somewhere useful. Spend ninety seconds writing what the form is for, who fills it in, what you will do with the answers, and which fields are non-negotiable. One good description beats five vague ones and gets you a better draft.

Edit rather than regenerate. Once you have a reasonable draft, changing a label, reordering fields or adjusting a validation rule is direct editing in the builder — no model call involved. Regenerating the whole form to change one question is the expensive way to do a cheap thing.

Duplicate instead of generating. If you need a second form that is largely the first with three differences, copy it. Agencies building the same intake form for successive clients get most of their leverage from a template plus edits, and it costs nothing.

What the AI is genuinely worth here

Worth separating two claims that get bundled together in marketing.

The generation step does not make a form good. It makes a first draft fast, and the first draft is the part people procrastinate on. What it produces still needs a human pass — the required-versus-optional decisions, the business rules only you know, the fields to cut. Generated field sets skew toward over-collection and toward free text where a dropdown belongs, which are the specific failure modes covered in what AI gets wrong when it writes form questions.

Where it is unambiguously strong is structure. A generated draft usually has better field types, more sensible validation and more coherent grouping than a non-specialist builds by hand, because those are pattern-matching problems. Take the structure, fix the judgement.

That division of labour is also the answer to whether it is cheaper to skip the AI and build manually. It is, if you value your hour at nothing.

A note on vocabulary

"AI runs", "credits" and "generations" mean roughly the same thing across the category, but the units are not comparable between vendors — one platform's run may cover an operation another splits into three. When comparing plans, translate to something concrete: how many complete forms can I build this month, and how many documents can I convert. Those two numbers are comparable; the raw allowance is not.

The same care is worth taking with the word "bot" in this category, which covers at least three different products — the terminology is untangled in what a bot form is.

Getting started

If you are choosing a plan, work out your monthly submissions and your storage need first, treat the run allowance as a sanity check, and remember that the month you build everything is not a typical month.

BotForms includes 100 AI-builder runs a month on Starter at $29, 1,000 on Pro at $79 and 5,000 on Business at $199, with extra runs available as an add-on at $8 per 500 and a 30-day trial on every plan. See pricing for current limits, features for what the generator does, or start a trial.