Curative cuts $600,000 Salesforce bill with AI-vibecoded CRM

Curative Inc., a health insurance startup, canceled its $600,000 annual Salesforce contract after using generative AI to build a replacement customer-relationship management system in two months, the company's chief executive officer said.

"Large language models let us write, test and deploy features faster than any off-the-shelf product could deliver," Fred Turner, CEO of Curative, said in an interview. "We went from zero to a fully functional CRM in 60 days."

The move comes as Gartner projects that agentic AI could expose as much as $234 billion in application spending to disintermediation by 2030, with AI agents completing tasks across multiple systems and reducing the need for traditional software interfaces. By the end of the decade, roughly 20 percent of enterprise SaaS spending will involve AI-driven interactions, the research firm estimates.

For Salesforce, which generated $34.9 billion in revenue in its latest fiscal year, the threat is existential. If even a fraction of enterprise customers follow Curative's lead, the subscription-based pricing model that has fueled the SaaS industry's growth could face structural pressure. Markets have already lost roughly $300 billion in SaaS valuations over the past 18 months on fears that AI agents will replace traditional tools, according to industry estimates.

The Curative case is among the most concrete examples yet of a phenomenon that technology analysts call disintermediation — the use of AI agents to bypass traditional enterprise software entirely. Rather than paying for a multi-module SaaS platform, companies can now describe their requirements to an AI model and receive a custom-built application in weeks.

"The SaaS apocalypse is overrated, but your use of the word disintermediation is correct," Shannon Kalvar, research director at IDC, said. "You can say, 'I need these things,' and Claude, ChatGPT, Perplexity, or whatever model you choose can go out and write a bunch of code and pull from capabilities. That ephemeral application becomes your new work surface."

For incumbent software providers, the response has been to emphasize what they describe as durable capabilities — governance, security, data management — that AI models alone cannot replicate. Workday CTO for EMEA Clare Hickie said the company is building a "front door to work" where employees use agentic services to ask natural-language questions about payroll and other issues, with answers drawn from enterprise data sources.

Freshworks CTO Murali Swaminathan argued that AI specialists such as OpenAI and Anthropic are unlikely to build the underlying systems of record that businesses require. "These AI specialists don't want to build everything," he said. "They want to be that layer where you engage, but the underlying layer will still be systems like ours."

Snowflake co-founder Benoît Dageville took a similar view, noting that his firm's data platform remains essential because "there is no AI without data." He said Amazon Web Services chose not to compete directly with Snowflake's data warehouse, instead building a partnership — a model he suggested AI giants may follow.

Still, the economics are hard to ignore. Curative's $600,000 annual saving represents the full cost of a Salesforce enterprise license for a company of its size. For larger enterprises with multi-million-dollar SaaS contracts, the potential savings from AI-built alternatives could run into the tens of millions.

Salesforce shares trade at roughly 25 times forward earnings, a premium that reflects the market's expectation of continued subscription revenue growth. If the Curative case proves replicable, that multiple could face compression as investors reassess the durability of SaaS revenue models. Gartner's estimate that $234 billion in application spending is exposed to agentic arbitrage suggests the risk is not theoretical.

This article is for informational purposes only and does not constitute investment advice.