On September 23, 2026, Ema raised a $77 million Series B to keep building teams of AI agents that run HR, IT and finance workflows inside enterprise systems. The number itself is not the story. The story is what Ema’s founder said about the software those agents sit on top of: large SaaS applications are becoming, in his words, “like a database.” For IT leaders, that line is worth more attention than the funding total. It describes a shift already underway in how enterprise software gets bought, used and priced, and it changes what a CIO should be negotiating this budget cycle. The raise also lands alongside market research pointing the same direction, which means this is no longer a story about one startup’s pitch deck. It is a data point in a pattern that should shape vendor conversations, renewal timing and budget requests.
The raise confirms a pattern, not a one-off
Ema’s business now has more than 50 active enterprise deals, over 1 million active users, and customers that include Google, Microsoft, PwC, KPMG and ADP, according to TechCrunch. Revenue has grown 50-fold over two years and gross margins sit around 80%, per the company’s own funding announcement. Those are the metrics of a company selling replacement, not augmentation. Founder Surojit Chatterjee told TechCrunch that customers are “already on the way to replace [large SaaS applications] completely, removing dependency on them.” That is a vendor describing his own customers’ procurement decisions, which makes it a more useful signal than most AI marketing copy. Net dollar retention around 180%, per the company’s own announcement, means existing customers are expanding their spend with Ema year over year, the same pattern that used to define a healthy SaaS renewal cycle. TechCrunch also cites Gartner projecting agentic AI could reach roughly 30% of enterprise application software revenue by 2035, up from about 2% today, which is the scale of shift a single funding round can only hint at.
The macro data backs up the anecdote
Ema is one data point, but it lines up with independent research. Gartner projects that task-specific AI agents will be embedded in 40% of enterprise applications by the end of 2026, up from less than 5% in 2025, a figure cited in UCToday’s coverage of the same Gartner research TechCrunch referenced. Separately, Gartner estimates that $234 billion in enterprise SaaS spending, roughly 20% of the total market, is at risk of disruption by agentic AI through 2030, according to CIO.com. That is not a prediction that every SaaS contract disappears next quarter. It is a statement that a fifth of current spend is now genuinely contestable, which is enough to change how any IT budget conversation should go this year.
Full replacement is still years away, and that matters for planning
It is worth being precise about the timeline, because overstating it leads to bad decisions. Deloitte’s own 2026 outlook, summarized alongside the Gartner figures, puts full enterprise-application replacement by agents at five years or more out, since incumbent platforms carry complex, deeply integrated workflows that are hard to unwind quickly. The near-term reality is narrower and more tactical: agents are displacing specific, repetitive, well-defined tasks and shrinking the number of paid seats a given SaaS contract requires, sometimes called seat compression, well before they displace entire platforms. That distinction should shape budget requests: this is a year for renegotiating and right-sizing existing SaaS contracts, not for ripping out core systems.
Pricing is shifting away from per-seat licensing
One reason this shift shows up in budget conversations before it shows up in ripped-out infrastructure is pricing. Deloitte, drawing on Gartner research in its 2026 technology predictions, projects that by 2030 at least 40% of enterprise SaaS spending will move away from traditional per-seat licensing toward usage-, agent-, or outcome-based pricing, according to Deloitte’s TMT predictions. That matters because once a workflow is run mostly by agents instead of licensed human seats, per-seat pricing stops matching how the software is actually used, and vendors know it. Expect renewal conversations this year to include new pricing tiers built around agent or API usage rather than logins, even for platforms that have no plans to be “replaced.”
What to check in your next SaaS contract
CIO.com’s coverage of the Gartner research offers a concrete checklist worth adopting directly. Read renewal contracts for clauses that restrict autonomous or API-driven use of the software, since some vendors limit or price differently when an AI agent, rather than a human, is the one clicking through the interface. Then check who owns the operational data and learning an agent accumulates while working inside that system. If that knowledge accrues to the vendor rather than to your organization, you are building a new form of lock-in on top of the old one. Finally, set governance rules now for where agents are allowed to act autonomously, what requires human sign-off, and how often those permissions get reviewed, rather than writing that policy after agents are already running in production.
Ema’s $77 million round is a funding story on the surface, but underneath it is a preview of the next few SaaS renewal cycles. The technology to compress seats and automate multi-step workflows already exists and is running in production at companies like Microsoft and PwC. The open question for most IT organizations is not whether to respond, but whether the contract, governance and data-ownership terms are in place before agents start doing the work, or after. Start the contract review this quarter.