askonor
AI-moderated interviews at the scale of a survey, with every theme grounded in a real participant quote. Built to support my own research and consulting work, then opened up as a standalone product.
Why I Built It
Most of the AI research tools landing in 2025 and 2026 are pointed at speed: faster transcripts, faster synthesis, more drafts. That’s useful, but it doesn’t address what I think is the actual bottleneck in research right now: trust.
If a researcher hands a stakeholder a finding without a defensible chain back to the source, the finding doesn’t move the roadmap. It’s another piece of content competing for attention with everything else the team is reading. AI tools that generate insights without grounding make this worse, not better. They produce more findings, faster, with weaker evidence.
I needed an interview tool for my own consulting work that didn’t have that problem. I built askonor to be the version that does it right, then opened it up so other researchers could use it too.
What It Does
askonor moderates structured AI interviews with your participants and returns themes, summaries, and the verbatim quotes that back every insight. You set up a study from your research goals in about 10 minutes, share a link with your cohort, and read the evidence when interviews close.
Researcher entry point: start from objectives, an existing guide, a standard format (NPS, CSAT, CES), or build from scratch.
Respondent view. NPS plus probing follow-ups. Same shape as a survey for the participant. Same depth as an interview for the researcher.
The Three Things That Matter
Grounded and traceable. Every theme is backed by a real quote from a real participant, with a one-click jump to the source. No hallucinated insights. No black-box claims. If you can’t show your stakeholder where a theme came from, the theme isn’t useful.
Live preview: every theme on the right traces back to a verbatim quote underneath it.
Same rigor, every interview. The AI moderator follows the same probing technique, bias rules, and topic boundaries on respondent #1 and respondent #100. No drift, no leading questions, no fatigue. Human moderators are good but inconsistent across a hundred sessions. AI moderators are consistent if you get the rules right, which is a research problem more than a model problem.
Pre-launch AI guide review. Flags leading questions, gaps, and ambiguity before the guide ever reaches a respondent.
Async at scale. Send a link. Participants interview themselves on their own time. You read the evidence when ready. Same shape as a survey for the participant. Same depth as an interview for the researcher.
What’s Live Today
- AI-moderated interviews with adaptive contextual probing
- Two modes: Exploratory and Survey (NPS, CSAT, CES, custom)
- Text chat on web and mobile
- Desktop browser-voice (OpenAI TTS plus browser mic; mobile auto-falls-back to text)
- AI guide generation from your research goals
- Pre-launch AI guide review (flags leading questions, gaps, ambiguity)
- Rules engine in every interview prompt (bias prevention, probing, topic boundaries)
- Researcher portal: projects, sessions, share links, cohort invitation tokens
- Open data: per-session export to JSON, CSV, or Google Sheets, drop straight into any analysis tool, BI dashboard, or database
- Enterprise-grade security, SOC 2-aligned controls
Coming Soon
- Interview quality scoring (depth, bias, coverage, missed-probe flags)
- Deterministic insight pipelines, fully traceable end-to-end
- Traceability UI: click any insight, jump to the source quotes
- Cross-session synthesis (themes across respondents, with audit trail)
- Runtime interview guardrails (rules enforced, not just instructed)
- Account-based auth: Google sign-in plus email/password
- Additional modes: passive listener (think-aloud), task-based usability
- Conversational voice: full-duplex (currently implemented but hidden)
- Direct integrations: Notion, Slack, BI connectors
How It Works
- Define a study from your goals.
- Review the AI-generated guide. The pre-launch review flags leading questions, gaps, and ambiguity before you send it out.
- Share a link with your cohort.
- Read the evidence when interviews close. Themes, summaries, verbatim quotes, all exportable.
Pricing
Usage-based on Interview Units (one IU equals roughly 5 minutes of interaction). Free trial of 3 lifetime sessions, no credit card. Tiered plans above that. Annual saves the equivalent of 3 months.
Status
Private beta. I’m using it in my own consulting work and running it with a small set of invited researchers. Not publicly available yet. If you’d like to try it or talk about what you’d want it to do, get in touch.
The Thread Behind It
This connects to a broader argument I’ve been making about where AI investment should be pointed in research workflows. Speed is the easy thing to optimize. Veracity (grounding, triangulation, traceability, fabrication detection) is the harder thing, and it’s the one that actually compounds. askonor is the version of that argument I can hand to another researcher and have them use.