@heyjunoco has been named among The Top 100 Innovators of 2025 by The @australian.
Building this with Michelle Gilmore and our team has been incredibly rewarding.
Turning decades of human research expertise into an AI agent that truly works isn’t easy, but our customers believe in the vision and push us to get better every day.
Juno is now conducting interviews 24/7—that real-world impact makes this recognition especially meaningful.
7/
You cannot claim "mostly positive sentiment towards AI in the workplace" when your interviewer is designed to generate positive sentiment.
@AnthropicAI has replaced the standards that good qualitative research depends on with an off-the-shelf assistant, and called it qualitative research at scale.
Credible research begins with interviewing quality.
This is the standard we should demand before trusting results.
#anthropic #claude #ai #moderator #LLM
@AnthropicAI just published a major study using @claudeai as an AI-interviewer.
They're calling it "qualitative research at scale."
We audited the transcripts. It scored 2.5/10.
Here's where it all fell apart. 🧵
6/
Why does this matter?
Weak interviewing → Weak data → Invalid conclusions.
When an interviewer reframes, flatters, and redirects, interviewees are influenced.
The resulting data is deeply flawed.
People make calls on bad data, and we all know where that leads.
“AI-moderated” research is having a moment.
@heyjunoco has been doing it since before the category had a label.
But not everything is what it seems. If it needs pre-written questions (“a discussion guide”), it’s not a moderated interview.
It’s a survey.
Deep insight comes from interviewing: exploring contradictions, knowing when to dig deeper, and pulling threads from earlier moments.
Surveys can’t do that. They stick to the script, use AI for a follow-up or two, then move to the next question. In short, they miss the gold.
One builds understanding. The other collects responses.
What would you choose?
@heyjunoco has been named among The Top 100 Innovators of 2025 by The @australian.
Building this with Michelle Gilmore and our team has been incredibly rewarding.
Turning decades of human research expertise into an AI agent that truly works isn’t easy, but our customers believe in the vision and push us to get better every day.
Juno is now conducting interviews 24/7—that real-world impact makes this recognition especially meaningful.
@coleplunges An automated system that has been afforded agency and decision-making power. It's trusted to do tasks toward an outcome, without supervision.
@robcurry_uk Try https://t.co/b2bpQa9MYx @robcurry_uk. AI-led interviews with real people. Much deeper data & insights. Surveys give you the what, not the why. Worth a look.
@uxdesigninst Nice read! Also, it's worth looking at https://t.co/b2bpQa9MYx – a unique way to collect qualitative data from real humans at scale, with actionable realtime insights