Nearly 1,900 people showed up to Quirk’s New York 2026 despite flash flood warnings across the city. Two days, packed rooms, one of the biggest shows Quirk’s has run. That turnout alone tells you something: quant research isn’t shrinking into a back-office function. People still want to be in the room for this conversation.
Now that the event has wrapped, aytm’s “Waves of Thinking” series and W5 Inc.’s recap give a real read on what got said on the floor. A few threads run through both, and they line up closely with the problems we spend our days solving.
1. AI’s Job Description Just Changed
aytm’s Elana Marmorstein opened the show pointing at a shift she saw building across sessions: less talk about what AI can technically do, more about how researchers apply it day to day, scaling qualitative research to quant-level speed, generating synthetic data, keeping a person in charge of the calls that matter. That’s a different conversation than the one this industry was having a year ago, when the question was still whether AI belonged in the workflow at all.
2. Data Quality Gets a Framework, and a Public Scoreboard
aytm’s Jonathan Goodbread, Head of Data Quality Strategy, laid out a four-part model on stage: prevent, protect, purify, prove. His company is going further than a slide, they’re committing to publish a quarterly report benchmarking their own data against the Insights Association’s GDQ standard, starting with data from the beginning of 2026.
That’s the same standard we build CodexMR’s platform around. Speed on its own doesn’t mean much if the dataset underneath falls apart the moment a client’s QA team opens it. Prevent, protect, purify, prove isn’t far from what our validation tools already do inside our own workflow: catching logic, routing, and scripting errors before they turn into a client-side fire drill instead of after.
3. Success Gets Redefined: Decisions Over Deliverables
Colgate-Palmolive’s Shourav Sen made a case worth sitting with: when AI gives a research team time back, the right move isn’t to produce more reports faster. The real opportunity is not to produce more reports but to spend more time understanding the business.
4. Real Behavior Still Beats Stated Intent
Bayer and Veylinx ran an in-home usage test built around an auction: testers spent their own money to buy the product before trying it. The result, nearly 80% were willing to pay more after using it than before. That’s a business case built on what people did with their own money, not what they said in a survey. It’s a small methodological choice with a large implication: the industry keeps finding that behavior is the harder, more useful signal to chase, even when it costs more to collect.
None of these four threads are about a single flashy demo. They’re about the same question from four directions: does the work hold up once someone checks it against reality.
That’s the question CodexMR’s platform is built to answer for the programming, validation, QA, and open-end coding stage of a study, faster, without the shortcuts that show up later as rework.
