π
Reply API V3 β Changelog
TL;DR: AI-drafted replies now surface in the Inbox API, companyDescription limits triple to 3,000 chars, and the AI SDR Pending Approvals API is fully documented ahead of its mid-June release. One breaking change β the AI voice LinkedIn endpoint drops playback-speed settings.
π’
New: live now
Inbox - AI Draft support
GET /v3/inbox/threads/{id} now returns AI reply data:
β’ aiDraft (nullable) - pending AI-drafted reply: message (HTML), cc, bcc, attachments. null if no active draft.
β’ aiSentType (nullable) on email & LinkedIn messages - aiGenerated / aiGeneratedEdited / autoSent.
Offers: larger company description
β’ companyDescription max length raised 1,000 β 3,000 on create + patch. icp / reasonForOutreach stay 1,000.
π Coming soon - targeted late-June 2026
AI SDR Pending Approvals - full API coverage
β’ GET /v3/ai-sdr/approvals β list (filters sequenceId / userId / channel / sortMode, top/skip)
β’ POST /v3/ai-sdr/approvals/send-bulk β batch of 1β20 pairs
β’ GET|POST|DELETE /v3/sequences/ai-sdr/{sequence_id}/contacts/{contact_id}/approval[/send|/regenerate]
β’ Regenerate is async: returns 202 {jobId} β poll GET /v3/background-jobs/{jobId} for subject/body.
β’ Reject discards the draft and removes the contact from the sequence.
AI SDR Sequences - limits & error contract
β’ companyDescription 1,000 β 3,000 on offer-generation inputs.
β’ Intent signals: technologyUsed & jobHiring now mutually exclusive.
β’ Email-only sequences reject contacts without an email address.
β οΈ
Breaking change
AI voice LinkedIn message - playback speed removed
Before: body accepted textToSpeechSettings with a speed multiplier.
After: textToSpeechSettings / speed removed; only text is accepted.
Action required: remove textToSpeechSettings from your request body - it has no effect.
π
Full docs: https://docs.reply.io/api-reference