Enterprise software has been fundamentally the same thing for 30 years: a database with a UI that employees click through. AI agents break that model entirely — and the $650B enterprise software market has no choice but to rebuild.
The Tool Era Is Ending
The dominant software paradigm since the 1990s has been: sell a tool, train humans to use it, charge by seat. Salesforce, Workday, ServiceNow — they all operate on this model. A human sits in front of the software and works. The software is passive. The human is active.
AI agents flip that. The software becomes the active party. It reads email threads, books meetings, writes first drafts, flags anomalies, routes tickets, updates CRM records, and escalates exceptions — all without a human clicking anything. The human reviews outputs and handles edge cases.
I've watched this shift accelerate dramatically over the past 18 months. The companies that figured out agentic workflows early — even in narrow domains like SDR outreach or contract review — are running circles around competitors still stuck in "human operates the tool" mode. The gap is widening faster than most enterprise buyers realize.
The Numbers Are Starting to Show Up
This isn't theoretical. Gartner projected that by 2028, 33% of enterprise software applications will include AI agents — up from less than 1% in 2024. That's not incremental. That's a category-level transformation compressed into four years.
The dollar flows are already moving. Microsoft's Copilot suite crossed $1B in annualized revenue in late 2025 — and that's before the agentic capabilities fully deployed. Salesforce's Agentforce, launched in late 2024, was their fastest product to $1B in pipeline generation. ServiceNow saw a 40%+ spike in deal size for contracts that included AI agents versus those that didn't.
Meanwhile, pure-play agentic startups are raising at multiples that make 2021 SAAS valuations look conservative. Glean raised at a $4.6B valuation in 2024. Writer hit unicorn status on the back of enterprise agent deployments. Hebbia and Harvey are both clearing $100M ARR milestones before most Series B companies have $10M ARR. The pricing power is real.
The Workflows Getting Disrupted First
- •Sales development: AI SDR agents are qualifying leads, personalizing outreach, and booking meetings at 10-20x the volume of a human rep for roughly 15% of the cost. Platforms like Artisan and 11x are already displacing headcount at growth-stage companies.
- •Legal and contract review: Harvey AI processes 10,000+ contracts per week for some of its enterprise clients. Law firms that used to charge $500/hour for junior associate contract review are watching that work vanish. The surviving firms are using Harvey, not competing with it.
- •IT and customer support: ServiceNow's agentic workflows are resolving 40-60% of tier-1 IT tickets without human involvement. For a 10,000-person company, that's millions of dollars in annual support cost eliminated.
- •Financial operations: AP automation, expense reconciliation, and audit preparation — tasks that previously required entire teams — are being compressed to a few agent-hours. Ramp reports customers saving 5+ hours per week per finance employee.
- •Software engineering: GitHub Copilot Workspace, Cursor, and Devin-style agents are already handling 20-40% of greenfield code tasks at companies that have deployed them seriously. The 10x engineer of 2026 isn't more talented — they have better agents.
What This Means for the Software Market
Here is the uncomfortable truth for legacy enterprise software vendors: their pricing models are built on seat count, and agents don't have seats. Salesforce charges per CRM user. If an AI agent is doing the work of five SDRs, does the company pay for five Salesforce licenses or one? The answer is neither — they'll pay for the agent platform that replaces the workflow entirely.
This is why every major SaaS incumbent is in existential transition right now. Salesforce didn't release Agentforce because they were excited about AI. They released it because they watched their renewal conversations get harder in 2024 and saw the same pattern I've seen in my portfolio: when a founder says "we cut our Salesforce spend by 60% because we rerouted outbound through an agent," that's not a niche experiment. That's a structural threat.
The opportunity for founders is enormous but narrow. Enterprise buyers want agents that own complete workflows end-to-end, not agents that are "AI-assisted" features glued onto existing tools. The winning companies will be the ones that define new workflow categories — not the ones that AI-wash existing product lines.
The question for every enterprise software company is no longer "how do we add AI?" It's "what do we do when AI replaces the reason customers bought from us in the first place?"
Stay current with VC and startup trends at Value Add VC. Originally published in the Trace Cohen newsletter.