Skip to main content

From respondents to an explorable panel.

HDTwin turns a target definition into an explorable synthetic panel through four steps.

  1. Build

    Create your panel

    Define the population you want to investigate or use existing research data to initialize and calibrate your Human Digital Twins.

  2. Ask

    Define the question

    Describe the topic, decision, product, proposition or scenario you want the panel to explore.

  3. Explore

    Run the discussion

    HDTwin orchestrates iterative surveys and simulated focus groups across the panel, combining structured questions with open-ended reasoning.

  4. Understand

    Analyse the results

    Explore responses, reasoning, opinion groups, disagreements and areas of convergence.

Overview - all 18 nodes

Scrollable region 1200 × 880 viewBox. On narrow screens scroll horizontally (drag, or focus the region and use the arrow keys).

From target definition to tuned consensusA target description and field interviews grow into aligned agents; the agents discuss a topic toward shared positions, checked against a human control group.Phase 1 · Define and collectPhase 2 · Build the modelPhase 3 · DeliberatePhase 4 · Check and tuneHermeneutic EngineNames the audience the study is about before any data is collected.Target definitionDefines the sample frame the virtual panel will represent.Panel definitionReal interviews and surveys supply the field evidence.Interviews and surveysTypical profiles and aggregated results summarise the fieldwork.Personas and survey resultsA shared map of meanings that keeps answers comparable across agents.Latent Cognitive ModelAgents adjusted to the character of the panel they represent.Aligned agentsChecks agent answers against the source material.Model reviewFlags answers that drift outside the source perimeter.Fit checkControls model bias by adapting models to each agent's character.Bias controlThe question the panel is asked to discuss.TopicThe panel members take part as anonymous discussants.Digital twinsEach agent proposes positions and questions the others.QuestionsAgents give reasons for the positions they hold.AnswersShared positions are grouped so agreement can be inspected.Consensus analysisOpinion groups show where the panel agrees and disagrees.Opinion groupsA small human group answers the same questions.Human control groupHuman answers give a reference the panel can be checked against.AnswersCompares panel answers with the human reference over time.Deviation analysis

The process steps in detail

  1. Target definition — Names the audience the study is about before any data is collected.
  2. Panel definition — Defines the sample frame the virtual panel will represent.
  3. Interviews and surveys — Real interviews and surveys supply the field evidence.
  4. Personas and survey results — Typical profiles and aggregated results summarise the fieldwork.
  5. Latent Cognitive Model — A shared map of meanings that keeps answers comparable across agents.
  6. Aligned agents — Agents adjusted to the character of the panel they represent.
  7. Model review — Checks agent answers against the source material.
  8. Fit check — Flags answers that drift outside the source perimeter.
  9. Bias control — Controls model bias by adapting models to each agent's character.
  10. Topic — The question the panel is asked to discuss.
  11. Digital twins — The panel members take part as anonymous discussants.
  12. Questions — Each agent proposes positions and questions the others.
  13. Answers — Agents give reasons for the positions they hold.
  14. Consensus analysis — Shared positions are grouped so agreement can be inspected.
  15. Opinion groups — Opinion groups show where the panel agrees and disagrees.
  16. Human control group — A small human group answers the same questions.
  17. Answers — Human answers give a reference the panel can be checked against.
  18. Deviation analysis — Compares panel answers with the human reference over time.
Walk through your research process

30-minute online conversation; bring a current research question and available evidence; next step is a process mapping session.