Agentic Commerce
For most of the past five years, B2B commerce conversations have centered on architecture. Headless or monolith. Composable or suite. Best-of-breed or suite-of-one. The assumption was that picking the right stack would set up the next phase of growth.
That assumption is no longer holding.
The B2B operators we advise have already made those decisions. Their platforms are modern. Their stacks are composable. Their integrations are in place. And in most cases the speed they expected hasn't materialized. New experiences take quarters to ship. Cross-team alignment slows everything down. AI initiatives stall in pilot.
The constraint has moved. The work is no longer selecting platforms. It is operating them.
We call this pattern the Operability Shift, and it's the most consequential development we've tracked in B2B commerce over the last 18 months. It shows up as three signals, each visible across BigCommerce's Commerce Live and B2B Online this year.
B2B buyers are increasingly finding products through AI assistants, LLM-driven search, syndicated marketplaces and retail media. The Future Proof Your B2B Playbook panel at B2B Online put a number on it: 26% of B2B search now happens through LLMs. Buyers are operating on channels suppliers don't control, evaluating options on surfaces operators don't own, and arriving at the storefront already shortlisted.
That breaks a foundational assumption. The storefront was where the buying journey converged. It's becoming one of several surfaces in a journey that converges elsewhere.
BigCommerce's positioning of Feedonomics at Commerce Live is one of the clearest platform-level responses to this we've seen. The PACSUN example made it concrete. Feedonomics handles catalog distribution across marketplaces, retail media and AI surfaces while BigCommerce powers the transaction layer. The storefront becomes a channel rather than a destination.
The same logic runs through the agentic stack being assembled around it. MCP for AI tool integration, Companion for operators, storefront agents for buyers, UCP for catalog distribution into AI surfaces. Those layers shorten the distance between discovery and transaction by making the catalog visible and actionable away from the storefront.
For operators, the implication is direct. Product data quality and feed strategy are infrastructure concerns now, not marketing concerns. Generic product descriptions get ignored by AI systems and buyers alike. Visibility depends on how clearly product data explains what makes a business different, and how cleanly it flows across surfaces.
Across the operators we advise, technology choices have largely been made. Stacks are modern. Platforms are capable. Integration patterns are in place. The friction slowing teams down is operational.
The patterns show up consistently. Complex internal stakeholder structures that delay decisions. Slow buying and approval cycles that stretch timelines. Disconnected workflows between marketing, merchandising, sales and IT. The persistent challenge of cross-functional alignment.
In that environment, additional architectural flexibility doesn't translate into additional speed. It usually adds more to manage.
The investments in Catalyst and Makeswift map directly onto this. Catalyst gives engineering teams a headless storefront foundation free of theming constraints. Makeswift lets operations teams edit pages and merchandising layouts without engineering tickets. Together they answer the operating model question explicitly. How does a platform team ship change without backing up the development queue?
B2B Online echoed the same theme from the operator side. Speed isn't a tooling problem anymore. It's a question of how quickly teams can align, execute and improve. Operators who invested in flexibility without investing in operating model design end up sitting on platforms they can't fully use.
Many teams that re-platformed in the last two years are operating below the capability ceiling of the platform they bought.
This is the most overlooked dynamic in B2B commerce right now. The next phase of work isn't another platform decision. It's the operating model that turns flexibility into pace.
Across our advisory work, the AI use cases that get funded and produce ROI in B2B share a shape. They compress known workflows. They remove manual steps. They make repeatable work faster. They don't invent new buyer experiences before fixing the workflows underneath.
BigCommerce's Purchase Order Agent is the cleanest platform-level example we've seen this year. POs in most B2B environments are still received, reviewed and processed manually. The work is repetitive, time-consuming and load-bearing for the rest of the order flow. The PO Agent doesn't reinvent the workflow. It removes the manual steps from one that already exists.
That pattern matters because most B2B AI conversations point the other way. The market narrative is focused on new buyer experiences, conversational commerce and agentic shopping. Those will matter eventually. But the operator value shows up first in operational efficiency. Invoice reconciliation, order processing, approval routing, returns triage, catalog enrichment, customer service deflection.
B2B buyers aren't asking for new ways to interact with suppliers. They're trying to complete known tasks more efficiently.
Operators chasing brand-new buyer experiences before fixing the operational workflows underneath are spending AI budget in the wrong order.
Reading the three signals together points to a different set of priorities than the architecture-era playbook.
Discovery readiness is a data and infrastructure investment. Feed strategy, catalog quality, AI-surface distribution. Platforms that can't extend the catalog beyond the storefront will lose share on surfaces operators don't control.
The operating model is the next investment area. Workflow design, cross-functional alignment, decision velocity, AI-assisted operations. Teams that don't invest here will sit on platform flexibility they can't fully use.
AI strategy should follow the workflow rather than the headline. Start where work already happens. Compress operational workflows before designing new buyer experiences. Operational ROI compounds and funds the next phase.
None of these are platform recommendations. They're operating model recommendations, and they apply across the major B2B commerce platforms. Platforms reading the shift correctly make the work more tractable, but the work itself belongs to the operator.
If you want to know where your own business sits against the three signals, the Operability Diagnostic scores twelve dimensions and gives you a tier and a next move in about fifteen minutes.
