Upgrade the plan — the only instant, self-serve fix
Sidebar → Account → Billing. A higher plan carries a larger daily allowance, and it applies immediately.
Troubleshooting
That is a hold message, not the AI's opinion.
When every visitor gets the same short line — “the team's a bit overloaded”, “leave your email” — the assistant is not answering at all. Something stopped it before it started, and it is nearly always the day's AI allowance. Open your dashboard: a bar across the top names the cause and the fix. If there is no bar, match the exact wording below.
Step one
Read the bar at the top of the dashboard
Sign in at app.orcalinq.com. If the AI is being held back, a coloured bar sits above every page. It is not dismissible on purpose: it reflects live state and clears itself the moment the cause is gone.
- Red bar. The AI has stopped answering. The bar says why and the button goes straight to the screen that fixes it.
- Amber bar. Still answering, but either most of the day's allowance is gone, or a countdown is running. Act today, not now.
- No bar at all. The daily allowance is not your problem. Go to the table below and match the exact sentence your visitors are getting.
The bar also shows today's usage — tokens used, tokens left, and a rough estimate of how long the rest of the day's allowance will last at the current rate.
Step two
Match the exact line your visitors see
| The line | What it really means | What clears it |
|---|---|---|
| “The team's a bit overloaded right now, but they want to hear from you. Leave your email and they'll reply within a few hours.” | The day's AI allowance for your account is spent. No AI answer is being generated at all. This is the classic “the assistant has gone monotonous” report. | Upgrade your plan, ask us to raise your daily cap, or wait for the reset at midnight UTC (5:30 am in India). |
| “Thanks for reaching out! I'll pass this to the team and someone will get back to you shortly — leave your email so they can reply.” | An Agency-plan account with no AI provider key, after the warning window has run out. The AI never runs. | Add a key under Provider keys — see the AI key page. It works again from the next message. |
| “I can't reach my answering service right now. Please try again in a moment, or use the Talk to a person button to bring in a teammate.” | Every model we could try failed on that message. Not a limit — an outage or a rejected provider key. | Try again in a minute. If it persists, check your own provider key is valid, then write to us with the time it happened. |
| “I don't have the exact details on that in your business's knowledge base, so I don't want to guess…” | The AI ran fine. Your knowledge base has nothing close enough to the question, so it refuses to invent an answer. | Add the missing content. Check the Behavior tab's auto-seed box and the Knowledge Base tab for that website. |
| “I'm your business's assistant, so I can only help with questions about your business…” | The question was judged unrelated to your business — or the site is in Funnel mode and the question is not on your allowed-topics list. | If the questions are legitimate, review the answering mode and the topic list. See answering modes. |
| “I understand your concern, and I want to make sure this gets proper attention. Let me connect you with a team member…” | The message was read as a complaint, so the assistant handed it to a person instead of answering. | Nothing to fix — but somebody has to be reachable. See chat modes. |
| One short acknowledgement, then silence. | The conversation is with a person, or the account is in WhatsApp only mode. The assistant stays quiet on purpose so it does not talk over your team. | Check the conversation in Live chat. In WhatsApp only mode, check the message actually arrived on the phone. |
Two of these lines are hold messages that only appear while daily limits are being enforced on your account. If you are seeing one of them, they are.
The numbers
What the daily allowance actually is
| Plan | Tokens per day | Very roughly |
|---|---|---|
| Basic (free) | 5,000 | about 5 conversations |
| Starter | 85,000 | about 85 conversations |
| Business | 300,000 | about 300 conversations |
| Pro | 850,000 | about 850 conversations |
| Agency | 2,000,000 | a soft ceiling; Agency reconciles separately |
- Tokens are not messages. Both the question and the answer count, and a long page of your own content quoted into an answer counts too. The conversation figures above are a rule of thumb, nothing more.
- Repeat questions are cheap. A question we have already answered is served from cache, so a busy day of similar enquiries costs far less than the raw count suggests.
- The free plan is for trying the product, not for running on. Five thousand tokens is enough to see the widget, the knowledge base and the hand-off working on your own content. If real customers are on your site, you want a paid plan.
- Reset is 00:00 UTC — 5:30 am in India. Nothing carries over, and nothing needs to be pressed.
Before it stops
The warning nobody notices: 80%
Past 80% of the day's allowance, answers switch to a lighter, cheaper model to stretch what is left. Nothing breaks, but the tone flattens and answers get shorter. If people tell you “the bot got worse this afternoon”, this is usually what happened — and the amber bar was showing the percentage all along.
This also happens when you have added your own AI provider key. Between 80% and 100% the lighter model is used regardless, because the daily counter is a platform-side limit rather than a billing one.
Fixes
How to get full answers back
Ask us to raise the daily cap on your current plan
The cap can be lifted per account, but only we can do it — there is no field for it in your dashboard. Write to support@orcalinq.com with your account and roughly how many conversations a day you expect.
Wait for the reset
If it is late in the day and the volume was a one-off, the allowance returns at midnight UTC by itself. Meanwhile visitors are still asked for an email address, so the enquiry is not lost.
Adding your own AI key: helpful, but not for this
Your own key moves the AI cost to your provider account and gives you a better model. It does not switch off the daily counter. Add one for quality and control, not as a way around the limit — see that page.
If that did not work
When the line keeps coming back
- Upgraded, and the hold message continues. The counter is per day and per account; a plan change raises the cap but does not reset the day's usage on its own. Check the bar's usage line — if used is still above the new cap, it clears at the reset.
- No bar, and none of the lines above match. Copy the exact sentence a visitor received and send it to us. The wording identifies the cause precisely; a paraphrase does not.
- It is only one website of several. The daily allowance is account-wide, so a per-site symptom is not the limit. Look at that site's answering mode and knowledge base instead.
- The line appears on WhatsApp or Telegram but not the website. Channels use the same AI and the same allowance, so an account-wide limit would show everywhere. A channel-only symptom is a channel problem — check that channel's connection.
- You are on the free plan and it happens every day by mid-morning. That is the free allowance working as designed. Five thousand tokens will not carry live traffic.
When you write in, include the time, the website, and the visitor's exact wording plus the assistant's exact reply. That is enough for us to name the gate without guessing.
One short line, one specific cause.
The bar in your dashboard names it. The table above decodes it. Neither needs a support ticket.