AI Agent
Suggest Mode vs Auto-Reply
Start in Suggest mode so agents review every draft, then graduate to Auto once you trust the bot's quality - or use Auto when offline as a middle ground.
Cove AI has four modes. The right one depends on how much you trust the bot’s current knowledge and how much coverage you need.
Change the mode in Cove AI - Setup - Mode.
The four modes
Suggest
The bot drafts a reply for every message it has an answer to. An agent reviews the draft in the inbox: accept it as-is, edit it, or reject it and reply manually.
The customer sees nothing until an agent approves. This is the safest starting point.
Start here. Running in Suggest mode for a week or two gives you real data on how good the drafts are before you let the bot reply on its own.
Auto when offline
The bot replies directly to customers only when your whole team is away (based on live agent availability and business hours). When at least one agent is online, it falls back to Suggest mode.
This is a good middle ground: full AI coverage during off-hours without giving up human review during business hours.
Auto
The bot always replies to customers directly, whenever it is confident enough. Agents see bot interactions in the inbox and can step in at any time.
Use this once you have seen consistently good drafts in Suggest mode and your training is solid.
Hybrid
Suggest when at least one agent is online, Auto when the team is offline. Combines the strengths of both: humans in the loop during the day, full coverage at night.
Graduating from Suggest to Auto
The AI agent analytics page shows a suggest-mode trust funnel: how often agents accepted, edited, or rejected the bot’s drafts.
When the acceptance rate (accepted + edited) reaches 80% or more across at least 20 interactions, Convot shows a nudge suggesting you try Auto when offline. That is the signal your training is ready.
If the acceptance rate is low, look at the Knowledge gaps report - it shows what the bot keeps failing to answer, so you know what to add to your knowledge base or Q&A snippets.
Auto mode safeguards
Even in Auto mode, the bot does not reply blindly:
- It still applies the confidence threshold. Below the threshold it escalates instead of answering.
- It respects the scope fence: any topic you list in Guardrails always goes to a human.
- It caps replies at the max-turns limit if you set one (after N bot replies, it hands off).
- It flags the conversation with “answered by AI” so agents can see the history.
Changing mode
- Go to Cove AI and select your app.
- Open Setup.
- Change the Mode dropdown and save. The new mode takes effect immediately for new conversations.
Existing open conversations continue with the mode that was active when they started.
Was this article helpful?
Thanks for your feedback!