{"version":"1.0","type":"card","id":"3d2157ed-644d-4238-94e1-3707a914d4d3","url":"https://stacklist.com/card/3d2157ed-644d-4238-94e1-3707a914d4d3","title":"(34) Building an outbound AI copywriter that isn't a black box. | LinkedIn","source_url":"https://www.linkedin.com/pulse/building-outbound-ai-copywriter-isnt-black-box-mitchell-keller-9xxpe/","note":null,"image":{"url":"https://ucarecdn.com/0de6df21-3fd0-4076-8eea-44f14848a307/","alt":"(34) Building an outbound AI copywriter that isn't a black box. | LinkedIn","width":1280,"height":720},"stack":{"id":"7c5a8832-3368-4e35-961e-f9273bcef8cc","title":"GTM Tools","url":"https://stacklist.com/c/business/stack/7c5a8832-3368-4e35-961e-f9273bcef8cc"},"created_at":"2026-09-17T20:42:09.855Z","updated_at":null,"aco":{"summary":"Building an effective AI copywriter requires heuristic-based copywriting skills, multi-level contextual data (account, person, company), and quality gates to prevent hallucinations. The system relies on feedback loops, ground truth testing, controlled experimentation, and a tech stack including Trigger.dev, vector databases, Supabase, Claude, and Langfuse to enable scalable 1-1 personalized outreach.","tags":["ai-copywriting","outbound-sales","personalization","prompt-engineering","abm","campaign-orchestration","vector-database"],"key_entities":[{"name":"Trigger.dev","type":"technology","confidence":0.95},{"name":"Supabase","type":"technology","confidence":0.95},{"name":"Claude","type":"technology","confidence":0.92},{"name":"Langfuse","type":"technology","confidence":0.9},{"name":"Firecrawl","type":"technology","confidence":0.85},{"name":"Railway","type":"technology","confidence":0.85},{"name":"Tailscale","type":"technology","confidence":0.85},{"name":"4o mini","type":"technology","confidence":0.88},{"name":"Jordan Crawford","type":"person","confidence":0.8},{"name":"Akshay Meena","type":"person","confidence":0.75},{"name":"Dayal Punjabi","type":"person","confidence":0.75},{"name":"account-based-marketing","type":"concept","confidence":0.92},{"name":"vector-database","type":"concept","confidence":0.9},{"name":"knowledge-graph","type":"concept","confidence":0.88},{"name":"feedback-loops","type":"concept","confidence":0.85},{"name":"LinkedIn","type":"organization","confidence":0.9}],"classification":"analysis","language":"en","confidence":0.85,"provenance":{"model":"claude-haiku-4-5","tool":"@stacklist/be@0.1.0","confidence":0.85,"timestamp":"2026-09-17T20:42:19.103Z"},"token_counts":{"approximate":1341,"cl100k":1073},"content_hash":"sha256:956adf64226994691d7431a80889ef330027b93228ba46ae0a78617c4ad38b97","acp_version":"0.2","body_available":true,"body_tokens":1341,"visibility":"public","agent_accessible":true,"status":"final"},"_links":{"self":"/api/public/card/3d2157ed-644d-4238-94e1-3707a914d4d3.json","html":"https://stacklist.com/card/3d2157ed-644d-4238-94e1-3707a914d4d3","md":"/api/public/card/3d2157ed-644d-4238-94e1-3707a914d4d3.md","stack_json":"/api/public/stack/7c5a8832-3368-4e35-961e-f9273bcef8cc.json"}}