You have run out of AI credits
MarqIQ has no AI credit allowance, so there is nothing of ours for you to run out of. What stops an AI run is one of four other things, and the message on screen names which. This article tells them apart, and says what happens to the part of a run that had already finished.
Before you start
Copy the exact wording of the message before you change anything. Each of the four causes below has its own sentence, and the sentence is what tells you which one you have.
There is no balance to check
No screen in MarqIQ shows an AI credit balance, because none exists. Plan & billing shows usage for two things only: people against your seat limit, and connected tools against your tool limit. MarqIQ does count how many Template Studio drafts your workspace makes each calendar month, in UTC, but nothing reads that count and nothing refuses a draft because of it.
Why there are no credits any more
There was an allowance: a monthly cap on Template Studio drafts, sold on the pricing page — 25 on a trial, 50 on Starter, 200 on Pro, 500 on Business. It existed because drafting ran on MarqIQ’s own AI keys, so MarqIQ was paying. Drafting now runs only on the AI accounts your workspace connects, so you pay your own provider directly and there is nothing of ours to meter. The allowance went on 18 September 2026 and no plan caps AI work by volume today.
One consequence is worth saying plainly: a message about credits that you see while enriching is your data provider’s message, not MarqIQ’s. AI enrichment runs on the accounts you connect, and the exact-match lookups run against vendors such as Clearbit, People Data Labs, Apollo.io, Hunter.io, FullContact and ZoomInfo with your key, drawing on that vendor’s own plan.
The four things that actually stop a run
“Connect an AI account to draft with AI. A workspace admin can connect one under Account settings → AI & Providers.”
Why: Template Studio drafting uses only the AI accounts this workspace connected, whatever else is set. With none connected there is nothing to call.
Fix: Open Account & Team, then AI & Providers, and connect one account. Several are listed with where to get a key and what it costs; Gemini’s free tier needs no card, and the card for it states the terms that apply where your team is.
“This workspace is set to use only its own AI provider keys, and none are connected”
Why: Somebody switched on Use only the AI accounts we own and no account is connected, so AI enrichment has nothing to run on. MarqIQ says so rather than silently doing nothing.
Fix: Connect an account under AI & Providers, or turn that setting off. With it off, enrichment still runs on the built-in heuristic and the exact-match lookups alone.
“Visibility checks are limited to 6 an hour per workspace on MarqIQ’s included AI engines”
Why: AI answer visibility can fall back to AI accounts of MarqIQ’s where you have not connected your own. Those cost MarqIQ nothing per token, but they are one quota shared by every workspace, so they carry fair-use limits. Your own accounts carry none.
Fix: Wait for the time the message gives, or connect your own AI accounts. The wording tells you which counter refused you: your workspace’s hour, your workspace’s day, or one of the two the whole pool shares.
“MarqIQ’s included AI engines are shared by every workspace, and not enough is left of today’s AI calls”
Why: Not your workspace’s doing. Every workspace together used the hour’s or the day’s allowance on the shared accounts. A large check may be refused in an hour where a smaller one would still fit, which is why the message names the size of yours.
Fix: Turn some prompts off and run a smaller check, wait for the time given, or connect your own accounts, where none of this applies.
“This run would ask 47 active prompts, and a run that uses MarqIQ’s included AI engines can ask up to 20”
Why: A run asks every active prompt on every engine. On the shared accounts that is capped at 20 prompts, and the run is refused before anything is counted rather than quietly cut down to a sample.
Fix: Turn off the number the message names on the Prompts tab, or connect your own accounts — a run on those asks every tracked prompt you hold, up to the 200 a workspace may keep.
“Too many generations — try again in a minute.”
Why: Template Studio drafting is limited to 20 generations a minute, counted per person rather than per workspace. Nothing else drafting does is limited.
Fix: Wait a minute. Where it keeps happening, something automated is calling it in a loop.
“The AI provider sent no draft back”, or enrichment returns nothing with an account connected
Why: Your own provider refused the call — an exhausted quota on their side, an expired key, or unpaid billing. MarqIQ passes their reason through rather than translating it.
Fix: Check the account with the provider directly, then re-verify the key under AI & Providers. The key is checked live when you connect it, so a key that saves is a key that worked at that moment.
The fair-use limits, in full
These apply only when a request would spend one of MarqIQ’s included AI accounts. A workspace running wholly on its own accounts is not limited at all.
| What you are doing | Per workspace, per hour | Per workspace, per day |
|---|---|---|
| Visibility checks | 6 | 16 |
| Prompt drafts | 10 | 20 |
| Recommendation drafts | 10 | 20 |
| Brief drafts | 30 | 40 |
| Brand profile analyses | 10 | 20 |
A day here is the UTC day. Above these sits a second set of limits across every workspace together, which is what the “shared by every workspace” message refers to.
Note
AI enrichment has one limit and it is a runaway-loop backstop rather than a budget: 120 runs a minute per workspace. Every provider it calls is your own account spending your own money.
What happens to a run that stopped partway
A visibility check asks every active prompt on every engine, which rarely fits in one request, so it spans several. Everything it stored is kept.
- Progress is read from what the run itself stored, never from a count the page sends, so a request that was stopped before storing anything leaves its calls pending rather than marking them done.
- The page shows the progress as calls — “12 of 40 calls done” — and continues from there. Closing the tab does not lose it; re-opening the check resumes it.
- A run counts against fair use once, when it starts. Continuing it is never counted again, so a long run cannot be refused halfway through for being long.
- A run is marked complete only once every prompt on every engine was attempted and the answers were scored. Until then no screen quotes its numbers.
- Only one check runs per workspace at a time. A second start within 75 seconds is told that one is already going, and told when a new one can begin.
Take care
Do not read a partial run as a result. The scores — mention rate, coverage, share of voice, sentiment — are computed over the answers that were scored, and a run that stopped early has fewer of them. The count is printed beside the rate for exactly this reason: read both or neither.
Removing the limits for good
Open Account & Team, then AI & Providers.
The page is admin-owned, because the same accounts are used by AI answer visibility, Template Studio drafting, the Suggested Plan brief and AI enrichment.
Connect an account for each engine you want measured.
Each row carries where to get the key, what the provider charges, and anything about their terms that matters — the key is verified live before it is stored, and it is never read back out to a browser.
Re-run the check that was refused.
A request that spends none of MarqIQ’s included accounts has no counters against it, so nothing can refuse it on fair use.
Decide whether to switch on Use only the AI accounts we own.
With it on, AI answer visibility and the Suggested Plan brief never fall back to an account of MarqIQ’s. It can reduce coverage, because an engine you have not connected is then not asked at all.
What that setting does not change
Template Studio drafting uses only the accounts connected there whether or not it is on. AI enrichment never uses an account of MarqIQ’s. The exact-match lookups are not AI and are unaffected.
If it did not work
The AI & Providers page offers no way to connect anything
Why: Your plan carries no feature that uses AI accounts, so the form is not shown.
Fix: Template Studio and the Suggested Plan brief arrive on Pro. A workspace that connected accounts on a higher plan and moved down keeps the form, so it can see and disconnect them.
You connected an account and the limit message still appears
Why: The limit counts a request that would spend any included account. A run asking an engine you have not connected still draws on the shared one.
Fix: Connect an account for every engine the run asks, or turn off the engines you do not pay for.
“An earlier request is still working on this check”
Why: A reload or a second tab dropped your browser’s request while the server carried on holding the workspace.
Fix: Wait the time the message gives. The same run continues from where it stopped; nothing was lost and nothing was counted twice.
You want to know how much drafting your team does
Why: The monthly draft count is recorded but is shown on no screen.
Fix: Read the usage on your AI provider’s own dashboard, which is where the money is spent.
Next steps
Related articles
Was this article helpful?