Y Combinator has developed a pattern. Each batch seems to surface clusters of startups chasing the same problem from slightly different angles—a sign, perhaps, of genuine market opportunity, or just founder groupthink. This cycle, wholesale distribution has become the focus, with a flurry of AI-driven entrants promising to automate the back-office drudgery that still defines much of what Whitespace claims is a $60 trillion global industry—a figure the company cites but that lacks independent verification.
Enter Whitespace, a London-based startup that emerged from YC's most recent summer cohort. The two-person team—yes, just two—claims to have built something they're calling an "AI Operating System" for distributors. Not a tool for one workflow. The whole operation: inventory planning, customer service, order processing, finance reconciliation, executive analytics. Their pitch is difficult to ignore, even if it invites skepticism: the company claims 80% productivity gains, three-month payback periods, and deployments live in two to four weeks—though these remain marketing assertions without third-party validation.
If the claims hold, it's an automation play with real teeth. If they don't, well, it wouldn't be the first time bold marketing arrived ahead of the product.
More Than Just Another Order Entry Tool
Whitespace isn't positioning itself as another point solution. Where competitors might tackle order entry or inventory forecasting in isolation, cofounders Alex Tung and Leon Yao are pitching something more expansive—a platform that spans multiple workflows and deploys what they call self-improving agents that learn from operator corrections over time.
The product works by plugging into a distributor's existing ERP, CRM, and email infrastructure to construct what Tung describes as a "living model" of the business. From there, agents take over discrete functions. Inventory agents handle demand planning, tracking seasonality and confidence intervals. They manage replenishment by monitoring supplier lead times and adjusting min/max thresholds dynamically. They flag slow-moving stock and suggest promotions or transfers between warehouses.
Sales and service agents field customer queries—order status checks, pricing questions, availability lookups—pulling data directly from the ERP rather than requiring a human to toggle between screens. Order operations agents process incoming orders from email, PDF, web portals, even fax (yes, fax still exists in this industry), and funnel them into the ERP. They learn from exceptions, generate purchase orders, reconcile supplier confirmations, and chase overdue line items.
Finance agents match invoices to purchase orders, track open accounts receivable, resolve disputes, and monitor customer credit risk. At the executive level, the platform generates weekly briefs on sales performance, margin erosion, stock health, and cash flow—then runs scenario models. What happens if a key supplier raises prices 8%? The system can model it.
The integration list is exhaustive: SAP, Oracle NetSuite, Epicor, Infor, Microsoft Dynamics, Sage, Salesforce, HubSpot, QuickBooks, Gmail, Outlook. Custom integrations available on request. Data is encrypted and not used to train models for other customers—a privacy stance that's table stakes in enterprise software now, but worth noting given how often AI vendors blur those lines.
Who's Building This?
Tung grew up around the family textile distribution business, so he didn't come to this market through a McKinsey deck. He was the first non-engineering hire at Ankar, an Index Ventures-backed startup, where he helped scale the company to seven-figure annual recurring revenue in a year. He spent time at BCG and studied economics at Cambridge—the kind of resume that signals analytical rigor but also raises the question of whether he's spent enough time in the trenches of a real distributor's chaos.
Leon Yao brings the technical credentials. A research engineer at InstaDeep, where he focused on AI for logistics and distribution, he's first-authored reinforcement learning papers and previously served as CTO at Test and Tutor, an edtech platform with over 1,000 users and partnerships with three schools. His academic background—MSc in machine learning, BEng in mechanical engineering from UCL—suggests he can build what Tung is selling.
Their thesis is simple, almost obvious: 30% to 50% of staff at distributors handle back-office work that hasn't fundamentally changed in decades. Email-to-ERP order entry? Still largely manual. Inventory planning? Spreadsheets and gut instinct. Customer service? A dance between multiple systems to answer basic questions. These workflows are repetitive enough to automate, Whitespace argues, but complex enough that single-purpose tools can't deliver compounding returns.
According to the company's website, typical deployments take two to four weeks. Agents initially operate in human-in-the-loop mode to calibrate to a customer's specific operations. Over time, accuracy "compounds" as the system learns from corrections. Pricing is usage-based, though no specific figures are public—a detail that might matter quite a bit to prospective customers trying to model ROI.
Early Customers, Limited Data
Whitespace lists two customers publicly. Kingfisher Direct, a UK distributor managing over 35,000 products across industrial and commercial categories, is using the platform to eliminate manual order processing and customer query workflows. The goal, according to a brief testimonial, is to grow revenue without adding headcount—a familiar refrain in automation pitches.
Warwick Fabrics, a UK textile and apparel wholesaler with global distribution, is deploying Whitespace for inventory planning and what the company calls "data unlocking." Both testimonials are short. Neither includes quantified results—no before-and-after labor hours, no hard savings figures, no improvement in stock turns or order accuracy. Just endorsements.
The company's website also displays logos of businesses its customers supply to—downstream relationships, not direct Whitespace customers: Amazon, Mercedes-Benz, Warner Bros., ExxonMobil, ASOS, Baxter, Fastenal, Jacobs, Johnson Matthey. This is a common marketing tactic that can mislead casual observers into thinking the startup has signed Fortune 500 accounts.
There's also a mock dashboard showing metrics like 99% availability versus a 97.5% target, a 38% reduction in dead stock, and a 6.2× improvement in stock turns. But these are presented as example metrics, not validated case data. In other words: illustrative, not proven.
A Suddenly Crowded Vertical
Whitespace is far from alone. The wholesale distribution AI category has become unexpectedly dense in recent months. Distro, which went through Y Combinator earlier this year, is building an AI co-pilot for sales reps at industrial distributors. Comena, also from YC, automates order processing for distributors and manufacturers. Stockline, another YC-backed entrant, is pitching an AI-native ERP specifically for food wholesalers.
Beyond the YC cohort, Lora markets itself as "the AI Operating System for Distributors" and has at least one live deployment at a chemical distributor. PhaseZero launched eight MCP-enabled AI agents for manufacturers and distributors in May. The "AI Operating System" label itself has become something of a catchphrase—VAST Data expanded its Cosmos ecosystem around similar positioning earlier this year, and PwC launched an enterprise AI agent operating system in late 2024.
It's unclear whether end customers care about the metaphor or just want measurable productivity gains. The branding arms race might matter more to venture capitalists than to procurement managers at mid-sized distributors.
What Whitespace claims as differentiation—at least in theory—is the breadth of workflows covered and the self-improving architecture. Where Comena focuses on order automation and Distro targets sales enablement, Whitespace is pitching a unified system that touches inventory, service, operations, and finance. Whether that scope becomes a strategic advantage or a distraction from nailing one workflow exceptionally well will depend entirely on execution.
The Market Opportunity (and the Risk)
Gartner forecast in April that supply chain management software with agentic AI will grow from under $2 billion in spend last year to $53 billion by 2030. That's a staggering market expansion, but it also means the category is still forming. Customers are experimenting, not standardizing. Whitespace is betting that distributors will prefer a platform approach over stitching together point solutions—and that its learning agents will deliver compounding value that justifies the inevitable switching costs.
The company's two-person team gives it agility. It also raises questions about how they'll scale support, sales, and engineering in parallel once they start signing customers at volume. Tung's post on the YC Launch page asked for introductions to wholesalers and distributors—a signal that the company is still in early customer development mode, perhaps more than the founders expected.
No funding details beyond Y Combinator's standard investment have been disclosed publicly. No independent press coverage exists. No case studies with hard numbers. No third-party validation of the 80%+ productivity claims. The product is live, customers are named, but the performance data remains entirely in-house.
For a vertical this large and this overdue for automation, Whitespace has identified the right problem. Whether it can deliver on the "operating system" vision faster than a half-dozen competitors—and prove the ROI in deployments that move beyond pilot projects—will determine if the bold positioning was prescient or just premature.
In wholesale distribution, margins are tight and patience for unproven software is limited. The window to prove value is narrow. Whitespace has made big claims. Now comes the harder part: making them stick.
