Bain & Company is out with a number that should make any SaaS founder sit up straighter. The consulting giant estimates that agentic AI could unlock a $100 billion market in the United States alone. This isn’t about replacing existing software platforms. It is about automating the messy, manual coordination work that happens between them.
The firm’s latest report, the second in a five-part series on software in the age of AI, digs into where this value actually lives. The core insight is refreshingly simple. Employees spend a huge amount of time shuttling data between enterprise systems. Think pulling a sales order from a CRM, checking inventory in an ERP, and then firing off an email for approval. That is the work Bain believes agentic AI can finally tackle at scale.
The Manual Glue Between Enterprise Systems
Rules-based automation and robotic process automation have been around for years. But they hit a wall when workflows involve ambiguity or information spread across multiple platforms. An invoice might come in as a PDF with inconsistent formatting. A support ticket could require checking a vendor management tool, an email thread, and a knowledge base. Traditional automation chokes on that level of variability.
Agentic AI, by contrast, can interpret unstructured messages, pull data from several sources, and decide on a course of action within policy guardrails. It can decide to approve a payment, escalate a ticket, or simply wait for more information. Bain argues that the market here is essentially converting labor-intensive coordination into software spending. And the opportunity is mostly untouched. The firm estimates vendors are already capturing only $4 billion to $6 billion of the U.S. market. That leaves more than 90 percent untapped.
The Global Picture and Where the Money Is
Outside the United States, the numbers get even bigger. Bain estimates that Canada, Europe, Australia, and New Zealand could add a roughly similar-sized market. Combined with the U.S., that brings the total addressable opportunity to about $200 billion. Not bad for a category that barely existed a couple of years ago.
The market is not evenly spread across business functions. Sales represents the biggest single slice at around $20 billion. That is less about automation potential and more about sheer headcount. Cost of goods sold and operations account for roughly $26 billion, driven by a large operational workforce where even modest automation rates translate into big dollars. Customer support, R&D and engineering, and finance each fall in the $6 billion to $12 billion range. These functions have both sizable workforces and higher automation potential in specific workflows.
Customer support and R&D stand out with the highest automation potential. Bain pegs roughly 40 to 60 percent of workflow tasks in those areas as automatable. They have structured data, standardized processes, and clearer output signals. Finance and human resources hover in the 35 to 45 percent range. Accounts payable and payroll are high potential; financial planning and employee relations, not so much. Sales and IT land at 30 to 40 percent, limited by relationship nuance, deal-by-deal variation, and unpredictable security incidents. Legal brings up the rear at 20 to 30 percent. Contract review and compliance are repeatable, but the consequences of errors demand tight oversight.
The Six Factors That Determine Automatability
Bain’s report identifies six factors that determine how much of a workflow an AI agent can realistically handle. Output verifiability is a big one. Workflows with clear verification signals are much easier to automate. Compiling code, reconciling invoices, and resolving support tickets are good examples. Work involving subjective judgment is far harder.
Consequence of failure is another critical factor. Workflows involving regulatory or financial risk require closer human supervision, even if an agent is technically capable. Tax filings, legal compliance, and security incident response fall into this category. Bain also flags digitized knowledge availability as a constraint. Agents need access to structured data and documented context. They also need machine-readable inputs, including decision logic that often sits informally with experienced employees. Integration complexity matters when workflows pass through several systems and APIs. Authentication layers and exception-handling processes add further complexity. The highest value areas are concentrated where no single system of record controls the full outcome. These workflows often span ERP, CRM, and support systems.
David Crawford, chairman of Bain’s global technology and telecommunications practice, puts it neatly. SaaS companies have spent two decades building positions around systems of record. The next source of advantage, he says, is cross-workflow decision context. That is the ability to interpret and act in workflows that move through multiple systems.
Real-World Proof and Adjacent Moves
The report cites several companies already riding this wave. Cursor, an AI coding tool, has surpassed $16.7 million in average monthly revenue after doubling in a single quarter. Sierra, an AI customer experience platform, has crossed $150 million in annual recurring revenue. Harvey, targeting legal workflows, passed $190 million. Glean, an enterprise search and AI assistant, hit $200 million. These are not hypothetical numbers. They are real traction in a market that Bain says is just getting started.
The report also points to GitHub as an example of a company using data from an existing core workflow to move into adjacent areas. It is a playbook that many SaaS companies are now studying closely. The question for founders is not whether agentic AI will reshape their industry. It is which side of that $100 billion equation they will end up on.
