{
  "title": "Building an outreach sequencer for managers who don't write copy",
  "slug": "outreach-sequencer",
  "description": "Cold email sat dormant across GoStudent's 100-plus tutoring centres. The missing layer was expertise, not tooling. Here is how we built the copywriter into a pipeline.",
  "archetype": "build-story",
  "solution_id": "gostudent-outreach-sequencer",
  "pillar": "workflow-automation",
  "reader": "technical-peer, readable by coo-ops-leader",
  "language": "en",
  "anonymisation_level": "L0",
  "publish_date": "",
  "canonical": "",
  "internal_links": [
    "semantic-layer"
  ],
  "key_takeaways": [
    "Sending tools assume a competent operator: they provide infrastructure, while the missing layer is usually expertise.",
    "Encode the expert as a pipeline: model research, one human gate where judgement matters, then a deterministic last mile for the step that must never be wrong.",
    "A reputationally dangerous model output gets a structural fence, not a politer prompt: hedged bands, schema-validated at two boundaries, surfaced only inside the approval gate.",
    "The model never writes sending-tool syntax; an injection function substitutes it and throws on unknown tokens, which turns zero wrong tokens into a contract instead of a hope."
  ],
  "sources": [
    "01-intake/examples/filled-intake-B.md"
  ],
  "visual_briefs": [
    "gostudent-sequencer-v1"
  ],
  "cta_variant": "plain-address",
  "opening_move": "what-we-built",
  "skeleton": "build-story",
  "register": "v2-engineering-essay",
  "target_query": "how to build an AI cold email tool for non-marketers",
  "author": "Antonio Marques, Brutal AI",
  "updated": "",
  "external_citations": [
    "https://arxiv.org/abs/2509.08514"
  ],
  "faq": [
    {
      "q": "Why not use the AI writer built into Smartlead or Instantly?",
      "a": "Those writers assume the operator can already judge audience strategy, structure and copy quality. GoStudent's centre managers could not, which was the actual gap. This tool owns strategy and drafting end to end, then exports to whatever sender each manager already uses."
    },
    {
      "q": "What does the human gate actually approve?",
      "a": "The persona research: audience description, buying context, the hedged affordability band, and tone. The approve button stays disabled until all four fields have been seen, every field is editable inline, and generation only runs from a persona a human has accepted or corrected."
    },
    {
      "q": "Could this pattern work outside tutoring?",
      "a": "The shape is not tutoring-specific: model research, one human gate where judgement matters, deterministic injection for the step that must never be wrong. Prompts, languages, audiences and sender-variable mappings live in configuration rather than code, so adapting it is mostly configuration work."
    }
  ]
}