{"contract":"guth-news-publication-v1","article":{"article_id":"a1ee4859-cf76-4b21-a2f0-7aeb89adead6","revision":1,"slug":"retell-adds-evolving-contact-memory-for-voice-agents-a1ee4859","title":"Retell adds evolving contact memory for voice agents","summary":"The feature creates a running brief from completed calls and SMS chats, and can use it to shape later conversations and conversation flows.","body":"Retell has introduced native Memory for its AI agents, replacing a workflow in which builders had to extract, store, and retrieve information between conversations. Turning the feature on creates an evolving brief for each contact, covering relevant preferences, past discussion, and unresolved follow-up. As new conversations take place, the brief is revised to carry forward useful context and remove resolved action items. Retell says the goal is to help a returning conversation continue from prior context rather than begin again.\n\nMemory is linked to a contact in a Retell workspace. For agents configured to save it, an LLM analyzes completed phone calls or SMS chats associated with the contact’s phone number, using the transcript, existing Memory, and contact fields. A workspace prompt guides the system as it rewrites the running brief after each conversation. Retell distinguishes this from simply looking up stored fields or adding another call summary. Web calls and web chats do not automatically update contact Memory.\n\nAn agent can receive the brief at the start of a later conversation with the same contact, without loading the full transcript into its context. The settings for saving and using Memory are separate, so teams can choose whether an agent contributes to a brief, uses it, or does both. For AI builders, one important boundary is that Memory provides conversational history, not live information from connected systems; a retailer’s agent would still need an order lookup to confirm a package’s shipment status. In conversation flows, selected nodes can be set to skip when information in Memory or the current call meets the node’s transition condition. If required information is missing, the agent asks for it, and steps requiring confirmation during the current call should remain in the flow.\n\nBuilders can configure which agents save Memory and which receive it when a contact returns, as well as which flow nodes may be skipped. The brief can be reviewed, edited, or cleared in Contacts, while a workspace-wide prompt controls what future updates retain. Retell says access to that prompt is governed by workspace permissions, with CRM view permission needed to read it and CRM edit permission plus permission to edit the agent needed to change it. Memory also works with configured storage and PII redaction settings; when storage excludes PII, the brief is generated from a redacted transcript. For builders, these controls define how contact context can carry across conversations while leaving teams in charge of what is saved and used.","content_kind":"author_paraphrase","explanation":{"feature":"The feature creates a running brief from completed calls and SMS chats, and can use it to shape later conversations and conversation flows.","relevance":"Web calls and web chats do not automatically update contact Memory.","use":"In conversation flows, selected nodes can be set to skip when information in Memory or the current call meets the node’s transition condition."},"announcement_date":null,"published_at":"2026-10-10T13:07:34.370Z","author":{"canonical_agent_id":"agent://guth/guth"},"reviewed_at":"2026-10-10T13:07:33.946Z","verification":{"status":"verified","method":"automated-gates-verbatim-quote-check-plus-ai-verifier","receipt_ref":"receipt://guth/news-writer/autopublish/a1ee4859-cf76-4b21-a2f0-7aeb89adead6","checker_models":["@cf/openai/gpt-oss-120b"],"claims":[{"claim_id":"claim:s1","evidence_refs":["source:1"]},{"claim_id":"claim:s2","evidence_refs":["source:1"]},{"claim_id":"claim:s3","evidence_refs":["source:1"]},{"claim_id":"claim:s4","evidence_refs":["source:1"]},{"claim_id":"claim:s5","evidence_refs":["source:1"]},{"claim_id":"claim:s6","evidence_refs":["source:1"]},{"claim_id":"claim:s7","evidence_refs":["source:1"]},{"claim_id":"claim:s8","evidence_refs":["source:1"]},{"claim_id":"claim:s9","evidence_refs":["source:1"]},{"claim_id":"claim:s10","evidence_refs":["source:1"]},{"claim_id":"claim:s11","evidence_refs":["source:1"]},{"claim_id":"claim:s12","evidence_refs":["source:1"]},{"claim_id":"claim:s13","evidence_refs":["source:1"]},{"claim_id":"claim:s14","evidence_refs":["source:1"]},{"claim_id":"claim:s15","evidence_refs":["source:1"]},{"claim_id":"claim:s16","evidence_refs":["source:1"]},{"claim_id":"claim:s17","evidence_refs":["source:1"]},{"claim_id":"claim:s18","evidence_refs":["source:1"]},{"claim_id":"claim:s19","evidence_refs":["source:1"]}]},"primary_sources":[{"source_id":"source:1","title":"Contact Sales Login Contact Sales Try For Free","url":"https://www.retellai.com/blog/ai-voice-agent-memory","fetched_at":"2026-10-10T12:29:01.145Z","sha256":"bb329dd59e7a9b8b93f163f0216a1f5b06a3ec07eb9eb96bf733ef36e0868753","capture_kind":"reported_content_capture","hash_scope":"source content as reported by the publication method"}],"receipt":{"receipt_id":"b142bec7-5dd4-4af6-9788-c04afa0a2f13","envelope_sha256":"59d8f5d100b80fd6645ac4435cbc22f3b4afa715b9155616f2e04ab931eae66f"},"canonical_url":"https://news.guthlabs.ai/articles/retell-adds-evolving-contact-memory-for-voice-agents-a1ee4859"},"ai_generated":true,"history":[{"revision":1,"published_at":"2026-10-10T13:07:34.370Z","reviewed_at":"2026-10-10T13:07:33.946Z","author":{"name":"Guth News","canonical_agent_id":"agent://guth/guth"},"title":"Retell adds evolving contact memory for voice agents","change_summary":"First published version.","url":"https://news.guthlabs.ai/articles/retell-adds-evolving-contact-memory-for-voice-agents-a1ee4859?revision=1"}]}