GetHeard Agentic

Four AI agents working together to take research from brief to report — each with memory, tools, and autonomous decision-making.

🗣️
Agent 1
Brief Agent
Conversational intake
🎨
Agent 2
Designer Agent
Generate + self-review questions
🎙️
Agent 3
Interview Agent
Contextual probing decisions
📊
Agent 4
Analysis Agent
4-pass transcript analysis
🗣️

Start a Research Brief

Chat with Alex, our research consultant agent. One question at a time — no forms, no dropdowns.

Open Brief Agent →
📁

View Projects

Browse projects created by the DesignerAgent, including question review history and quality metadata.

View Projects →
📊

Agentic Reports

Reports generated by the 4-pass AnalysisAgent, with per-transcript extraction and cross-transcript synthesis.

View Reports →

What changed vs. the classic version

Classic (v1) — One-shot LLM calls

  • Web form to enter research brief
  • Single Gemini call generates all questions
  • No quality review of questions
  • State machine decides when to probe
  • Single Gemini call generates full report
  • No self-critique of analysis

Agentic (v2) — Tool-calling loops

  • BriefAgent chats with client to collect brief
  • DesignerAgent generates then self-reviews questions
  • Iterative revision until quality bar is met
  • InterviewAgent decides to probe based on answer quality
  • AnalysisAgent runs 4 passes: extract → synthesize → write → critique
  • Self-critique patches weak sections before finalising