Bain Says Cross-Workflow AI Is a New Competitive Layer, $100B Market Cited

For two decades, the pitch in enterprise software was simple: own the authoritative record of what happened, and the relationship that came with it would follow. Salesforce tracked every sale. Workday tracked every employee. SAP tracked every factory floor. The company that held the record held the pricing power.

Bain & Company thinks that era is ending. On May 7, 2026, the firm published research arguing that a new competitive layer has emerged — what it calls cross-workflow decision context: the ability of an AI agent to see, interpret, and act across ERP, CRM, billing, and support systems simultaneously, without waiting for a human to reconcile what one system says against another. The Boston consulting firm estimates this new market could be worth $100 billion in the US alone, and roughly $200 billion when Canada, Europe, Australia, and New Zealand are included. The firms already building the layer — Cursor, Sierra, Harvey, Glean — are posting ARR numbers that suggest the market is not theoretical. But the consulting report is also a sales document, and the numbers warrant scrutiny.

"The new competitive advantage is what we call cross-workflow decision context," David Crawford, the chairman of Bain's technology, media, and telecommunications practice, said in the research post. "For two decades, SaaS companies built competitive moats around systems of record. Agentic AI reopens this contest."

The framing is an era argument, not a product launch. The question is whether the era argument holds.

Cross-workflow decision context is Bain's term for a capability that is recognizable in practice: an AI that can check a customer's billing tier, cross-reference an open support ticket, pull their contract history from the CRM, and decide what to do without a human routing the information. That is structurally different from the first generation of SaaS tools, which stored data but required a person to read it and act. The new layer does not just record. It decides.

The automation potential varies by function in ways that matter for competitive strategy. Customer support and research and development sit at the high end, with roughly 40% to 60% of workflow tasks automatable today, according to Bain's estimates. Legal is high in repeatability but low in tolerance for error — contract review is mechanical, but a bad output has severe consequences, keeping legal automation in the 20% to 30% range. Sales and IT sit at 30% to 40%. Finance and HR fall in the 35% to 45% band. The spread is not a technical limitation. It is a business design question: which workflow you automate first, and who you let make that decision, is itself a source of leverage.

The companies already building across those workflows are attracting the most concentrated investor attention of the current enterprise software cycle. Sierra, which automates cross-enterprise customer support resolution, crossed $150 million in annual recurring revenue and raised $950 million in May 2026. Cursor, an AI coding assistant, has surpassed $2 billion in ARR — doubling in a single quarter, according to Bain's research. Harvey, which chains multiple AI models to handle legal workflows, passed $190 million in ARR. Glean, an enterprise search platform, hit $200 million. These are not pilot numbers. They are enterprises signing annual contracts and renewing them.

Cursor's trajectory is the sharpest data point in the Bain report. The company went from $100 million in ARR in January 2025 to $2 billion by February 2026 — eighteen months. Whether that trajectory is replicable or an outlier is the central question for every builder deciding where to place a bet right now.

The skeptic's case is in the open. Consulting reports are written to get on the agenda of the executives who hire the firm. The $100 billion US market estimate is a top-down assertion, not a bottom-up survey of buyer behavior. Bain puts current vendor capture at $4 billion to $6 billion, with more than 90% of the opportunity still uncaptured — a figure that is unfalsifiable by design. The ARR numbers, cited from company disclosures, reflect firms with Bain relationships rather than a random sample of the market. Bain did not publish its methodology for the market sizing in the public-facing post.

What the numbers cannot tell you is which companies will still be here in three years, or whether the incumbents simply build agentic overlays on top of their existing databases and keep the relationship. Salesforce has announced Agentforce, an autonomous AI layer across its CRM. Moveworks has been competing in enterprise support automation. Whether these constitute a genuine new moat or a feature update on top of an existing system is the unresolved question that will determine whether Bain's $100 billion market materializes — and who captures it.

The system-of-record war produced Salesforce, Workday, and SAP. The system-of-action war is still being decided. The customer has stopped asking whether to automate. The question now is which layer gets to be the brain.