How it works

Sentiment + signal

Pranan reads tone shift across the thread. Heated language flags an escalation.

Suggested response, not draft

On flagged threads, Pranan offers talking points instead of a full draft.

Escape valve always

Manual override per thread. You're the founder, not Pranan.

When the AI says 'you should write this one'

Three signal types trigger escalation flags. First: sentiment shift. If the thread tone moves from neutral to charged, the AI flags. Second: explicit risk language. 'Considering canceling,' 'need to escalate,' 'this is unacceptable' all trigger. Third: stakeholder change. If a senior person joins a thread that was previously between juniors, the AI flags for context.

Flagged threads surface in the morning brief without a draft. Instead, the AI provides talking points and tone guidance. The user writes the actual reply.

This is intentional. The AI's value on most threads is volume. On critical threads, the AI's value is preparation, not delegation.

Try escalation detection free

300 drafts a month free. Voice fingerprint from your sent mail in under 5 minutes.