The rapid rise of autonomous AI agents is forcing CRM vendors to confront a problem that extends beyond adding new AI features. Many legacy platforms were designed for human workflows, not AI agents capable of making decisions, coordinating tasks, and acting across enterprise systems.
Some CRM vendors are rebuilding around AI-native architectures that unify data, context, and automation to support both human and agentic workflows. Others are layering AI features onto existing platforms, fueling debate over whether legacy CRM architectures can keep pace.
Software-as-a-service (SaaS) vendors are also facing pressure to rethink traditional per-seat licensing as autonomous AI agents increasingly perform work once handled by human users.
Jason Eubanks, CEO of Aurasell, an AI-native go-to-market (GTM) platform, argues that many legacy CRM vendors are attempting to layer AI agents onto outdated infrastructure originally built for human-led workflows. Those platforms were never designed for autonomous decision-making, real-time orchestration, or AI-native workflows. He sees a quiet tension brewing within the enterprise.
Companies are embracing the promise of an agentic future, only to realize that their underlying systems and data environments were never built to support AI agents. As a result, many enterprises are reevaluating their long-term dependence on legacy CRM platforms.
AI Tests the SaaS Model
Marc Benioff, co-founder, chairman and CEO of Salesforce, recently pushed back on predictions of a SaaS apocalypse, arguing that AI makes SaaS better rather than replacing it. Eubanks thinks Benioff is only half right.
According to research firm Forrester, the so-called SaaSpocalypse is not about CRM platforms abandoning the cloud for on-premises hardware. Rather, it is the threat AI agents pose to the established per-seat SaaS subscription model. Traditional SaaS CRM is typically built around per-seat licenses costing $50 to $300 per user per month, manual data entry, and rigid user interfaces.
That prospect could force CRM vendors to rethink how they design, price, and monetize their platforms.
"What people are tempted to do is take a legacy platform like HubSpot or Salesforce and put a cloud connector on it and say now AI has access to my CRM data. Well, that's great. But your CRM data shows poor hygiene, incomplete updates, and account conflicts. It's a mess," Eubanks told CRM Buyer.
Building CRM for AI Agents
Aurasell is one example of an AI-native platform built around that philosophy. Rather than replacing Salesforce, Eubanks said customers are increasingly adding AI orchestration layers above their existing CRM platforms.
According to Eubanks, the GTM stack has undergone little meaningful change in the past 20 years. He wanted to build a platform that incorporated AI from day one.
Since launching in August 2025, Aurasell says it has grown to more than 50 customers, with 37% replacing HubSpot or Salesforce. According to the company, Aurasell generates more than $6 million in annual recurring revenue, powers more than 185 million agentic workflows, and serves hundreds of users daily.
On June 24, Aurasell introduced Agent Builder, a no-code platform that lets GTM teams create AI-driven workflows using natural language. It combines structured and unstructured customer data to coordinate actions across sales and marketing systems. Agent Builder analyzes customer signals in real time to automate workflows while supporting enterprise governance and role-based access controls.
"We’re at an inflection point where the question is no longer whether AI belongs in GTM, but whether your stack was built to support it," Eubanks offered. "Legacy tools weren’t built for autonomous action. Agent Builder is. That’s not incremental improvement but a bet on what GTM software becomes."
Early Customer Results
Jim Robshaw, chief data and AI officer at Xerox, noted that his company moved 85,000 accounts into Aurasell to improve the digital halo of each account, leveraging the thousands of Aurasell agents on the platform.
"We had to create an AI-enabled competitive takeout sequence where the platform determines the next best action across all 85,000 accounts. Aurasell has enabled both of those business process challenges within weeks of implementation," he said.
TCI Transportation connected a third-party intent signal in just 10 minutes without a single line of code or RevOps involvement, Ross Calame, executive vice president of sales, said. Agent Builder then created the downstream workflow, identifying accounts, routing records, generating tasks, triggering outreach sequences, and surfacing visibility to the top-of-funnel team.
"Agent Builder didn't just connect the data. It automatically built the agentic workflow around it, helping us operationalize the signal from day one. It wasn't a one-shot integration. It was a dead shot," Calame said.
Marc Manara, head of startups at OpenAI, added that Aurasell is showing what is possible when OpenAI models are grounded in the way GTM teams actually work.
"Their new Agent Builder combines the planning and tool-use capabilities of our latest models with Aurasell’s deep GTM context and verification layer, helping teams build complex workflows designed to execute reliably," he said.
Those early deployments illustrate why some observers believe AI-native platforms could change not only CRM architecture but also the economics of enterprise software.
Why AI Challenges the CRM Business Model
Even if predictions of a SaaS apocalypse prove overstated, AI is accelerating software innovation while challenging the economics of traditional seat-based CRM licensing.
Most AI capabilities in CRM are delivered through cloud-based SaaS platforms because generative AI and large language models require substantial computing resources, ongoing model updates, and scalable infrastructure.
Private cloud and on-premises deployments represent a much smaller share of enterprise AI implementations, and legacy CRM systems typically rely on middleware and APIs to connect with cloud-based AI services rather than providing built-in generative AI capabilities.
As AI agents replace work traditionally performed by teams of human users, enterprises may require far fewer CRM seats. That reality challenges the economics of per-user subscriptions and encourages consumption- or outcome-based pricing instead.
Eubanks argued that legacy CRM platforms were built on architectures that assume humans enter data into static, structured data models.
"The reality is, in an AI era, you have to have a unified data architecture that supports unstructured and semi-structured data, along with structured data," he said. "A platform like ours is built to contemplate agentic workloads as well as human workloads, living side by side harmoniously."
CRM Still Has a Data Problem
Not everyone believes CRM's biggest challenge is its technical architecture. Some industry observers argue the larger issue is how organizations collect and use customer data.
Kathy Baldwin is a business systems strategist and founder of Finally Business Systems, an AI-driven financial and accounting platform for small and medium-sized businesses. She says too many organizations focus on their offer stack, their visibility, and their objectives. They fail to map the buyer journey and align it with their offers to find their gaps, friction, and bottlenecks.
"They keep throwing more time, money, and resources without going to the root cause," she told CRM Buyer.
Baldwin added that modern CRM platforms have accumulated so many tools and so much data. But companies are not creating buyer/client profiles from the breadcrumbs left by customer engagement, habits, and purchases.
"Our clients are telling us by how they engage, and too many companies have not successfully captured the actual behavior of the buyer and systematized it," she said.
Whether enterprises modernize existing CRM platforms or adopt AI-native alternatives, experts agree success will depend as much on data quality and workflow design as on the AI models themselves.
Loading ...
