Kerry Lu has been crisscrossing the country lately, racking up 25,000 miles in four weeks, according to his LinkedIn. His pitch? That market research—the often glacial process of surveying panels, waiting for responses, analyzing results—can happen in minutes instead of weeks.
It sounds like hyperbole. Yet Lu and his co-founders at Auxos, a Y Combinator-backed startup based in San Francisco, insist they've cracked it. Their solution involves something that would have seemed like science fiction not long ago: creating AI-powered digital twins of actual customers, then running unlimited tests against them.
The concept is deceptively simple. Define your ideal customer profile. Auxos recruits real people who match it, interviews them, then builds what the company calls 1:1 AI replicas—each digital twin corresponding to an individual respondent. Assemble a panel of these twins (the startup claims it can scale to over 1,000), and suddenly you can test ad creative variants, pricing tiers, product concepts, landing page copy—whatever you want—without waiting for live humans to respond. The twins don't just spit out quantitative preferences; they explain why they chose what they chose.
When Weeks Become Minutes
On Auxos's website, the value proposition appears in stark relief: a comparison table. Traditional research panels? Two to six weeks to results. Auxos? Minutes. Traditional panels struggle with niche audiences. Auxos claims instant availability. Conventional testing cycles limit variant volume. Auxos says it handles ten-plus simultaneously.
For B2B SaaS founders and marketing leaders—the company's target audience—that speed differential isn't just convenient. It's potentially transformative. These are teams operating on tight budgets and tighter timelines, where waiting a month for survey results can mean missing a product launch window or blowing through ad spend on untested messaging. "Simulate every decision before it reaches a real customer," the company's tagline promises.
Whether that's realistic is another question entirely.
The Unusual Trio Behind It
Auxos went live in mid-May 2026, though pinning down exact dates for early-stage startups always proves tricky. What's clearer is the founding team's pedigree—and their oddly specific shared background.
Ashton Daniel, the CEO, worked at Boston Consulting Group on agentic AI deployments while also running surveys, panels, and interviews. Kerry Lu, CTO, came from Amazon Ads infrastructure, where he reportedly built ad simulation systems modeling a significant chunk of U.S. e-commerce traffic. Jerry Wu, the chief product officer, spent time at Meta and Instagram building infrastructure for Reels, ads, and the Facebook feed—work the company credits with driving over $100 million in incremental annual revenue.
All three studied computer science at top-tier universities: Columbia, Duke, and UPenn. And here's where it gets specific—all three were NCAA fencers and Team USA members. It's the kind of detail that might seem trivial (and probably is), but it's also the kind of founder synergy story accelerators love. Shared discipline, literal and metaphorical.
The company remains small. Y Combinator materials listed three team members initially; LinkedIn suggests somewhere between two and ten employees, the vagueness typical of startups still in stealth mode. Yet the founders have been moving. Lu's travel spree. Wu taking Auxos to The Lead Summit in New York in late May, working booth 206A. Daniel hosting dinners for marketing leaders. The hustle is real, even if the headcount isn't.
Behind the Curtain
The Auxos workflow breaks into three phases: Align, Build, Deploy. First, the company collaborates with clients to define their ideal customer profile. Then it recruits matching real people and interviews them. Finally, it constructs the digital twins—one per respondent—and clients can start querying their synthetic panel.
A UI screenshot on the website shows "Twin sync · 3 of 200 live," hinting at how the platform tracks synchronization as twins come online. Use cases displayed include packaging tests, ad creative comparisons, product concepts, pricing evaluations. One example counter reads "Ad creative A/B 2,104," though what that number represents isn't immediately clear, and the company hasn't published methodology or validation studies publicly.
Results supposedly arrive both quantitative and qualitative. Drill into individuals or segment-level insights. Learn not just what customers prefer, but why. At least in theory.
A Suddenly Crowded Field
Auxos is hardly alone in this space—perhaps less alone than its founders might prefer.
Brox announced just weeks before Auxos that it had built 60,000 digital twins and secured strategic funding, claiming tenfold revenue growth, according to VentureBeat. Quantilope launched something called "Category Twins" for synthetic research around the same period. Even Qualtrics, the incumbent survey platform, now offers Qualtrics Edge for synthetic responses at scale.
Then there's the parade of earlier-stage entrants: Deepsona, AnalyzeUS, SynthPanel, YourFocusGroup. Each pitching some variation on synthetic audiences. The market research community itself remains deeply divided. Scroll through recent Reddit threads and you'll find professionals debating whether digital twins represent a legitimate tool or just the latest overhyped buzzword.
As of mid-2026, Auxos hasn't yet published benchmarks. No backtesting studies. No human-versus-twin concordance metrics. Pricing remains undisclosed—the website funnels visitors straight to "Get a demo." There are no customer logos, no testimonials visible. Security certifications like SOC 2 or ISO compliance? Not mentioned on the public site.
For a product this early, that's not necessarily damning. But it does raise questions.
The Bet on Speed Over Perfection
How accurate are these twins, really? How do they handle edge cases—customers whose preferences shift week to week, or rapidly evolving sentiment around a brand? What happens when the real people the twins are modeled on change their minds?
Auxos isn't claiming to replace all traditional research. Not yet, anyway. But for teams needing directional signals fast—marketers who want to test ten messaging variants before committing media budgets, product teams operating in niches where conventional panels move like molasses—the promise holds obvious appeal.
The company is making a specific bet: that in today's market, speed trumps perfection for a growing share of research budgets. That founders would rather have decent insights now than perfect insights in six weeks, when the window has already closed.
Whether the twins prove reliable enough to justify that wager? That's the experiment Auxos is running—on itself as much as its customers.
