Agent design patterns
Eight ways agent products lose trust
These are eight failures that recur across agent products. For each one, the table below shows what it looks like, what it costs, what to do instead, and the interaction pattern that addresses it.
| What it looks like | What it costs | Instead | Addressed by |
|---|---|---|---|
| Confirm only at the end The agent works for twenty minutes and then asks whether it got it right | One error in step three means redoing all twelve | Place checkpoints deliberately, weighted early | Scheduled confirmation |
| Transparency theatre A long list of tool calls labeled as reasoning | There is too much to read and people stop checking the work | Summary by default, detail on request | Altitude control |
| One autonomy switch for everything A single on or off covering every task | After one bad experience people will just turn the whole feature off | One dial per task type | Autonomy dial |
| Silent irreversible action It already happened, with no preview and no undo | People rarely trust the agent again after this | Preview, undo, and an honest expiry | Intent preview |
| Notification flood Several agents each interrupting on their own schedule | People stop reading, including the one that mattered | One attention budget across every agent | Interruption budget |
| Chat as the container for everything Parallel and overnight work forced through one transcript | It is hard to tell what the agents are doing right now | A decision queue and a run board | Agent inbox |
| Anonymous fleets Agents with no name, owner or remit | It is hard to trace who authorized each action | Owner, scope, spend cap, review date | Agent registry |
| Confident guessing Ambiguity resolved quietly into something plausible | The wrong answer looks finished and easy to believe | Three ways out: notify, ask, review | Escalation triad |