Included resolutions per plan
Plan | Resolutions per month |
|---|---|
Starter | 25 |
Pro | 50 |
Scale | 100 |
A trial includes 50 resolutions for the whole trial. Legacy Business and Enterprise plans include 100 a month, and the free plan includes none.
A resolution is a conversation the AI agent answers on its own. Follow-ups in the same conversation count once. A conversation handed to your team counts as zero.
When an answer counts as resolved
The agent's answer counts as a resolution in three cases.
The customer confirms it worked. When the customer writes something like "works now" or "found it", the agent marks the thread done. The resolution counts right away. A plain "thanks" for an answer the customer hasn't tried yet doesn't count on its own.
The customer goes quiet for 24 hours. When nobody writes for 24 hours after the agent's answer, the conversation counts as resolved. This applies on every channel. A short "thanks" after the answer still counts as quiet.
A teammate marks the thread done. Productlane reviews the finished conversation right away. If the agent solved it and the teammate only checked in, it counts as resolved. If a teammate was needed, it stays with your team and counts as zero.
In live chat, Productlane also runs this review when a teammate wrote in and the chat then stays quiet for 24 hours. See Live Chat.
When the customer comes back
On email, Slack, Microsoft Teams, and live chat, a customer may write again within 72 hours with a real follow-up. Productlane then takes the resolution back, so it doesn't bill. A bare "thanks" leaves the resolution in place.
Productlane takes a resolution back only within the calendar month (UTC) it counted in. Once that month is invoiced, the resolution stays.
Following resolutions in a thread
Each resolution shows in the thread's timeline. Your team also gets a Resolved by the agent inbox notification.
The thread's Details panel shows a badge for where the agent's answer stands:
Resolved by the agent.
A teammate was needed.
Pending, with the time it counts as resolved if the customer stays quiet.
The customer replied after the agent's answer.
When the agent stops responding
The usual cause is a workspace that has used its included resolutions for the month.
With pay-per-use off, the agent stops resolving new conversations and hands them to your team until the allowance resets next month. A visitor who opens the widget sees, in their own language: "Our AI assistant is paused right now. Talk with a human and we'll pick this up from here."
With pay-per-use on, the agent keeps resolving conversations at $0.79 per resolution past the allowance, up to any monthly cap you set.
To check and fix it:
Go to Settings > AI > Usage.
Compare your usage with your plan's allowance.
If you have hit it, turn on pay-per-use.
Pay-per-use and the resolution cap
With pay-per-use on, resolutions past your allowance cost $0.79 each. Turning it on sets a starting monthly cap.
Click Adjust cap on the usage page to change it, or Remove cap in the editor for no limit. Your usage and your cap show together on the usage page.
The monthly AI budget
Every paid AI feature draws from one monthly budget: resolutions, docs generation, translation, changelog generation, close loop follow-ups, and more.
With pay-per-use on, your included plan allowance is used first across all of these features. Pay-per-use billing only starts once that allowance runs out, and the usage page lists included usage separately from usage above the allowance.
The Monthly overage limit on the usage page caps that budget in dollars. Click Set a limit to turn it on, starting at $50, or Adjust limit to change it. Remove limit in the editor turns it off again, so overage runs without a cap.
When AI pauses for the month, a banner on the usage page says why:
During a trial, AI is paused for this month asks for a payment method to lift the limit right away.
Past your overage limit, AI is paused at your overage limit shows the amount spent and the day credits refresh, with Adjust limit to raise it.
Clicking Generate Changelog while AI is paused shows the same message, with a link to the usage page. A changelog already generating in the background stops, leaves the entry unchanged, and shows "Changelog generation failed" with a Try again action.
