Scot Wingo doesn’t need a long introduction. ChannelAdvisor (Now Rithum) founder. Took it public in 2013. Sold to PE in 2022. Founder of North Carolinas Tweener Fund which has invested in 167 companies across the Research Triangle (19 have exited!). And now, at a moment when most serial founders would be deep into their third act of comfortable board seats and golf, he’s back in the arena building ReFiBuy — a bet on agentic commerce as the next generation of how people shop online.
We sat down with Scot at ShopTalk and covered a lot of ground: what ReFiBuy actually does, why the SaaS moat playbook is breaking, how early-stage founders should think about survival and consolidation right now, and where agentic commerce goes in the next twelve months.
Here’s the full breakdown.
What ReFiBuy Actually Does
The core insight starts with a data asymmetry problem.
For twenty years, brands and retailers operated in what Scot calls “keyword jail.” Google gave you four words of shopper intent. That was the gold standard. And because you only had four words, you only needed to provide basic product information — size, color, the fundamentals.
That world is over.
Today’s AI engines know an enormous amount about the shopper — preferences, purchase history, dietary restrictions, shoe size, lifestyle. But they know almost nothing meaningful about the products themselves. Most product catalogs are still basically spreadsheets with a handful of attributes, and there’s no standardization across them. Just the word “small” is described in over 400 different ways across the industry.
ReFiBuy fixes the product side of that equation. They help brands and retailers expand their product catalog attributes, layer in rich Q&A content that addresses shopper concerns and occasions, and incorporate review data in a way that LLMs can actually use. The goal: give the AI engine enough context about a product that it can make a confident, accurate recommendation to a shopper who’s already told the engine everything about themselves.
The same enriched catalog payload that works for ChatGPT, Perplexity, Copilot, Meta, and Gemini also works for on-site LLM search — Rufus on Amazon, Sparky on Walmart. One infrastructure investment, multiple distribution channels. That’s the bet.
The ReFiBuy Team and Why It Came Together
Scot didn’t start fresh. He went back to a problem he couldn’t solve during the ChannelAdvisor years — the canonicalization and taxonomy mapping challenge that sits underneath all of this product data work — and asked whether agentic AI frameworks could crack it now.
Turns out they can.
The engineering team includes Cameron Bo and James Frawley, both ChannelAdvisor veterans, along with Derek Conlin on go-to-market. Part of the team is drawn from a ChannelAdvisor office in Limerick, Ireland — a hotbed of canonicalization talent that traces back to Dell’s internationalization work at the University of Limerick’s computer science program. It’s a niche skill set, and Scot’s been cultivating it for twenty years.
The Buy Box for Inorganic Growth
Scot’s current thinking on acquisition is more nuanced than most operators at his stage.
Engineering talent isn’t his constraint — he has it. The traditional SaaS acquisition for GTM talent is tempting, but he’s skeptical. Anyone who built a substantial go-to-market motion before 2022 is working off a playbook that’s increasingly broken. You’d be acquiring methodology debt alongside the customer base.
What he’d actually pay for:
Customers and revenue streams from companies whose GTM is sputtering but whose install base is real. Convert ten percent of a long-tail customer list into your model and you’ve got a meaningful ROI even if you write off everything else.
Audience. The most controversial item on his buy box — and the most forward-looking. In an environment where noise is at a nine out of ten and climbing, a founder or company with a loyal, engaged audience is a genuine strategic asset. Scot openly says he wouldn’t have said this five years ago. He’s saying it now.
The underlying logic: inbound is working at ReFiBuy in a way outbound SDR motions never could in this environment. His Substack, Retail Agentic, is on track to drive half of lead generation. Content is the new cold call, and audiences are the new distribution.
Sharks in the Water: What Scot Tells Struggling Founders
Scot published a piece warning early-stage founders about the current environment. The message, synthesized:
Buy runway first. Your existing investors are your best option. Your number one job as CEO is to not run out of money. Everything else is secondary.
Cut costs relentlessly. Not selectively. Relentlessly.
Diagnose your churn data. If you’re a B2B SaaS company selling to mid-market or enterprise and you’re not seeing churn creep up, look harder. It’s probably there. When you find it, trace it back — it’s most likely a signal that your competitive moat has been eroded by AI-native alternatives, not that your product got worse.
Retool go-to-market. The outbound SDR motion is trending toward zero efficacy. If your pipeline depends on it, you need a plan B. The noise level is too high and buyers have tuned it out almost completely.
Early-Stage Consolidation: The Down-Market M&A Opportunity
One of the most interesting parts of the conversation was Scot’s framework for startup-to-startup combinations — something he walks portfolio companies through regularly.
The logic is straightforward: two companies with complementary assets (one has proprietary data, one has distribution; one has product-market fit, one has runway) can be stronger together than either is alone. Combined back office, combined capital, combined time. In a market where lead investors are pulling term sheets at the eleventh hour and FUD is driving weird behavior on all sides, buying yourself more runway through a combination isn’t giving up — it’s smart capital allocation.
The challenge is always valuation. Tech founders default to the metric that makes them look best — last raise, revenue multiple, EBITDA multiple — and they avoid the money conversation until it’s almost too late. Scot’s prescription: get to the economics conversation early, use a simple one-page MOU framework to force the issue, and build a basic model that shows the combined thesis. The acquirer builds the model. The acquirer sells the target on why a smaller piece of something real is better than a larger piece of something dying.
Cash on balance sheet, Christian noted, is increasingly a force multiplier in deal structure — especially in AI-adjacent deals where capital raised is being treated almost like a proxy for validation. Christian cited the Goldcast/Cvent deal ($300M acquisition on roughly $8-10M ARR, with $35-40M raised) as an illustration of the dynamic. The old revenue multiple framework is being supplemented, and sometimes replaced, by a capital-raised multiple in high-conviction AI categories.
The Twelve Moats Framework
When evaluating early-stage companies — either as an investor or as an operator thinking about defensive positioning — Scot uses a twelve-factor framework for AI-era competitive moats. He built it from research across a16z (Alex Rampell’s talk is worth finding), NFX, and a handful of other VC frameworks.
The most defensible moats, in his view:
Proprietary data that can’t be synthesized in parallel. The mythical man-month problem applied to data — nine women can’t make a baby in a month. If your data advantage comes from iterative customer feedback loops and workflow embedding over time, it can’t be replicated by a well-funded competitor throwing engineers at it. That’s a real moat.
Workflow embeddedness. Get deeply enough into a customer’s operational workflow and the switching cost becomes structural, not just contractual. The best companies do both simultaneously — they’re embedded in the workflow AND the workflow runs on proprietary data they’ve been building for years.
Founder-market fit. For the earliest stage, the jockey matters as much as the horse. You want founders who deeply understand the market, stay agile, and are thinking hard about go-to-market — not just product. The best mousetrap in the world is worthless if nobody can find it.
Twelve-Month Predictions on Agentic Commerce
Scot publishes an annual predictions list on the Jason and Scot show — ten years of predictions, tracked annually. His calibration note: in the old era he was always a year early. Now he’s pulling predictions in by a factor of three because the pace of change has accelerated that dramatically.
A few of his 2025/26 predictions have already come true. Notable ones:
Facebook entering agentic commerce — happened. A “super protocol” that can talk to MCP and other lower-level protocols — UCP arrived. And the big one: by this holiday season, he’s predicting ten percent of e-commerce transactions will route through agentic commerce in some form.
The underlying thesis: filtered navigation is a broken experience. Conversational commerce — whatever you call it — gets shoppers to answers faster, with more personalization, and with less friction. E-commerce has been growing in line with retail for years. Agentic commerce has the potential to re-accelerate the gap by making the digital experience meaningfully better than the physical one again.
Scot Wingo is the founder of ReFiBuy and ChannelAdvisor (IPO 2013, taken private by Insight Partners 2022). He writes the Retail Agentic Substack and co-hosts the long-running Jason and Scot Show. He’s an LP in 167 companies through the Tweener Fund, focused on the Research Triangle Park ecosystem.
Subscribe to In/Organic for weekly M&A and startup coverage across agency and SaaS.
