Synthetic-Respondent Research Agents: Electric Twin Turns Market Research Into a Single-Prompt Autonomous Project

Synthetic-Respondent Research Agents: Electric Twin Turns Market Research Into a Single-Prompt Autonomous Project

A new vertical entrant for the market map (Market Map & Positioning: Where Does Hey, Lefty Fit?): on October 12, 2026, UK-based Electric Twin launched an autonomous research agent on top of its synthetic-audience platform — a distinct data-provenance layer where the corpus is a simulation (built from a brand's own audience data, AI/ML, and social-science method) rather than web retrieval.

What the agent does. Per IT Brief's launch coverage: the tool "produces report-ready findings in less than 30 minutes" from a single prompt, and is designed "to run research tasks such as focus groups, follow-up questioning and further lines of enquiry without requiring users to draft a survey themselves." The agent autonomously decides "which tools to use, which audiences and questions to select, and which results to explore in more detail," can "ask similar survey questions in different ways to test whether wording is leading, and run adversarial testing to verify outputs." It targets both research professionals (throughput) and "non-specialist teams in areas such as marketing, product and strategy."

The headline validation claim. Electric Twin "tested the agent against a UK government study into public perceptions of hosepipe bans in England. It says the original study took seven months to complete, while its tool reproduced the work in 25 minutes," surfacing "the same headline findings" plus additional insights — and argues the comparison used post-training-cutoff data, so "the results could not have been learned in advance." The accuracy claim follows the now-familiar self-graded pattern: "It claims the system reaches up to 96% accuracy, though the announcement included no independent verification¹" (IT Brief).

Business model: bundling, not metering. The agent "is available as part of its standard contract and will not carry an extra charge for existing or new customers" — a contrast with the per-run/per-request meters spreading through the rest of the research-agent market (Exa Agent Ultra's per-run budget, Parallel's CPM, Perplexity's per-invocation tools). Existing customers named: News UK, Sage, Centre for Net Zero, Virgin. Backers include Marc Andreessen and Atomico ("recently raised millions").

CEO Alex Cooper's framing of the value prop: "In market research you used to have to think of a question, then wait weeks before you got your results and even longer for an in-depth research project. Now, you can ask and find out in minutes whether you're onto something. Our new agent removes the last barrier which is preventing analysis from being at every team's disposal. It means you no longer need to know how to design a survey to get a rigorous, data-backed answer to a question about your customers."

Why it matters. (1) It extends the vertical-research pattern into market research/insights — previously uncovered in the map — where the moat is proprietary audience data plus simulated respondents, not search indexing². (2) The agent owns the research design itself (audience selection, question construction, adversarial verification), pushing further up the autonomy stack than report-writing deep-research tools. (3) Its unverified 96% claim and self-run validation mirror the benchmark-credibility problem tracked in Exa Agent Ultra: Deep Research Becomes a Metered Subagent-Swarm API — With Self-Graded Benchmarks and The Research-Agent Measurement Layer: Arena's $200M Independent-Eval Business Meets Exa's ATLAS Benchmark — the audit-grade vertical (The Audit-Grade Vertical: Research Agents That Sell "Show Your Work" — Lumivero, Menos AI, HITZ) is the obvious counter-positioning.