Now that code is cheap, how does learning keep up? A live demo of the answer.
From concept to confidence with Dscout AI | Config 2026 · Jenny Shear
19 min total·Actually worth watching closely: ~4 min·3 must-watch clips
- 0:12 – 2:30Listen
Code got cheap; research got more valuable
Figma opens the session and hands off to the Dscout team, and both speakers take the stage. The talk's core tension is on the table immediately: if code is nearly free, why not skip research and just ship?
Your ability to iterate is capped not by build speed but by the quality and speed of feedback. The real problem is getting learning to keep up with building.
Mostly two people talking on stage and introducing each other, with nothing on screen you need to watch. Listening while doing something else is enough.▶ Jump to 0:12 - 2:30 – 5:00Listen
Don't get high off your own supply
A pointed observation: when you build with AI, the tool keeps telling you your direction is right and your ideas are brilliant, and it's very easy to get high off your own supply. Mistake that flattery for real signal and you slowly lose your read on what actually lands.
Craft isn't only making things masterfully, it's the art of resonating deeply, and resonance can only be tuned against real users.
Pure argument, with at most a few large words on a slide. You won't miss anything with your eyes closed, and it's easier to follow the reasoning that way.▶ Jump to 2:30 - 5:00 – 7:14Listen
Everyone can make; who sets direction
Now that we're all makers, the number of decision-makers has jumped, and plenty of companies sit on a pile of prototypes and proofs of concept they can't move forward. The question shifts from how to build faster to what is worth building, and team alignment becomes the new bottleneck.
A POC glut isn't a capacity problem, it's a direction problem; what's missing is a shared basis for judgment.
Same argumentative rhythm as the previous stretch, with the slides in service of the talking. Follow the line of reasoning by ear.▶ Jump to 5:00 - 7:14 – 11:28Skim
From finding fit to keeping up with flow
A replacement for product-market fit: products fall in and out of fit so fast it feels like minutes, so rather than chasing a sweet spot that holds for years, the skill is being agile and moving with the flow of the market. Then the AI Studio announcement begins with the first principle, that any builder on the team can run research.
Give it your learning objectives and a link to a Figma prototype or site, and the AI uses that for context and walks you, with UX expertise built in, through every decision a study you can trust requires. What blocks non-researchers was never desire, it's time and confidence.
This stretch runs on concept diagrams and product screenshots. A glance gets the point, so you can scrub through and pause when a product UI comes up.▶ Jump to 7:14 - 11:32 – 14:20Skim
Guardrails beat approval processes
Once research is open to everyone, quality becomes the question. The answer given here is to bake the organization's best practices straight into templates and make them the required vehicle for research, replacing unmaintained documents, Notion hubs, and layers of review.
A template isn't there to give you a starting point, it's a guardrail that stays out of the way. Governance belongs inside the workflow, not bolted on beside it.
Mostly template configuration screens and flow diagrams. Skim the screenshots for the structure; no need to listen line by line.▶ Jump to 11:32 - 14:20 – 16:15Watch
AI moderators, and talking to your data
AI-moderated interviews: the AI decides on its own where to probe deeper and which thread to follow, and sessions run by the hundred at once across time zones and languages. Then a look at asking questions of your data directly, and the automatically curated collection of key moments.
The counterintuitive part is that participants are more open and candid with an AI moderator, which is what finally makes interview-grade qualitative data scalable.
The section closes with a scrolling wall of highlights AI pulled from many sessions. Whether the picks are good, and whether they'd save you hours of watching recordings, is a judgment only watching can give you.▶ Jump to 14:20 - 16:15 – 18:36Watch
A full loop inside Figma Make
The connector announced in beta that day puts the whole feedback flow where the design work is already happening: call Dscout from the Figma Make chat, build a launchable usability test, then bring the collected feedback back in and have the prototype revised on it.
The value of the loop isn't a few fewer clicks, it's that no tool switch sits between testing and iterating, which is what lets the pace of learning match the new pace of building.
This is step-by-step work in the real interface: what's typed, what gets generated, how the prototype changes only makes sense on screen, and narration alone can't tell you whether it works. The last stretch is booth directions, so you can stop there.▶ Jump to 16:15