You’ve no doubt heard of the SaaSpocalypse.
The US health insurance company Curative is a firm believer. Its Co-Founder and CEO Fred Turner made headlines when he decribed how his company built its own Salesforce alternative in just two months using AI.
This has cut 80% of their SaaS bill this year alone.
If you’re an executive who wonders how much your employees really use the tools you’re paying for, stories like this no doubt fire the imagination.
Where once building custom solutions for in-house problems simply wasn’t possible, today this option is back. A single employee, working over a couple of days, can even throw together a working prototype that appears to prove it.
So why does anyone continue to pay a third-party company year-after-year for CRM?
As the Founder and CEO of a tech solutions and recruiting firm, I will say it’s not the straightforward answer it was just five years ago.
AI is radically accelerating development, expanding what teams can do, and making many “crazy ideas” suddenly very attainable.
Yet, all this said, many CIOs are also finding the devil’s very much in the details.
Today I look a little deeper at stories like Curative’s. Is the time really here where companies can ditch their SaaS and build (and own) exactly what they need?
Is Salesforce worth it in 2026? The king vs AI CRM solutions for enterprises
Curative’s story is not unique.
Every month I hear about another (still mostly small to medium-sized companies) that’s shifted from subscription software to bespoke solutions they’ve both made in-house and solely own.
Here are some examples of very small companies profiled by The Information that cut their software costs from anywhere between 40 and 80% per year in this way:
- Atlanta-based real-estate investment company Greenleaf Management used Claude Code and Replit to build their own solution to move off multiple SaaS contracts. The 55-person firm reportedly now spends just $300 a month on maintenance and saves about $100,000 per year. In addition to Salesforce, they’ve also moved off real-estate software Entrata and Yardi.
- Utah-based startup Atonom is around the same size (by employee count), and replaced a $40,000 Salesforce contract with a custom-made CRM (using Lovable). They anticipate this will cost just $1,200 per year, but their chief revenue officer noted they have modest needs. “No one was using Salesforce to its full potential.”
- You maybe haven’t heard of the Seattle Seawolves, the professional rugby team (with 70 employees). But you have now. They used Claude Code to move off both Salesforce and the AXS ticketing system. Like Curative, the turnaround was quick, building their solution in just four months. Their owner reported saving around $100,000 per year in software costs, with revenue up 25% since the start of the season.
A common refrain in many of these cases is that Salesforce is massive, complex, and extremely powerful. It may be beyond what a lot of smaller companies need, and one of the exciting things about AI is that they can build and maintain their own answer to numerous problems that would once have been impractical.
But practicality—especially over time—is key because as Curative’s CEO went on to explain, his AI costs have shot up alongside maintenance needs, which he called “definitely one of the most challenging pieces” so far in going it alone.
Still, these changes aren’t limited to 50-person companies or the one-offs like Curative.
French pharmaceutical firm Sanofi is of a different scale (75,000 employees), and it’s reportedly cut 80% of its ServiceNow use by routing work through agents (built with Claude Code and Cursor).
Their anticipated savings?
$10 million per year.
What is the ROI of Salesforce compared to building a custom CRM?
Salesforce ROI is the right place to start, because it’s one thing to build a new system and another to maintain, secure, and scale it over time.
In considering Salesforce vs custom CRMs (or broader utilities that incorporate CRM), workflows are critical. Are your needs narrow, and highly defined? Does your Salesforce licensing include capabilities you don’t use, or use to any real benefits?
The next question is capacity. AI may vibe code you an impressive-looking solution, but as most businesses are now all too familiar with, it’s one thing to make a dazzling AI pilot and another to make a lasting, robust, maintainable, and secure solution.
Doing the latter most certainly requires strong technical resources to help you go beyond attractive interfaces and ensure effective, secure, and properly governed data management and AI connections on the back-end.
Where I see Salesforce still bringing higher returns is through depth, breadth, and safety.
The last point may raise eyebrows after reports of a widespread hacking campaign that targeted the Salesforce platforms of some very large companies for a long-running stretch of last year. We wrote a series of articles on this, and the vulnerability had little to do with Salesforce, with the attackers often tricking employees into setting up malicious connections (or downloading malicious tools).
And while using a large, industry-leading solution may open the door to such attacks because it has well-known surfaces, in truth Salesforce remains extremely reliable for safety and compliance.
As the company told Business Insider’s Ben Shimkus, Salesforce still has some 150,000 companies among its clients, and among these are many that specifically trust them for navigating complex industry regulations like HIPAA.
When making a Salesforce AI CRM financial comparison, you have to consider:
- Initial investment, or licensing and implementing Salesforce successfully versus custom AI development. (And yes, PTP does provide Salesforce consulting services as well as custom AI development and integration.)
- Ongoing operating costs. This is usually the focus of these stories: AI tokens vs Salesforce subscriptions. While Curative’s CEO has admitted their Anthropic charges have grown by a multiple of six every month (from thousands to millions of dollars), this includes far more than their CRM replacement. And while it’s likely harder to measure cleanly, I suspect in the short-term this remains a big win for AI for many companies. But over time, this becomes less clear when considering ongoing patching, system integrations and updates, and scaling.
- Business value returned. What are the real revenue, sales cycle, customer satisfaction, customer retention, service cost, and employee productivity numbers?
- Risk tolerance. Our latest PTP Report AI roundup is full of security issues, from OpenAI to Anthropic to Microsoft, and while AI-built solutions can of course be secured, considerations must include the impacts from potential compliance failures, downtime, technical debt incurred to handle unexpected edge-cases or failures. Here the win most certainly goes back in the Salesforce column.
The Curative case is intriguing because they were able to cut a $600,000 annual Salesforce bill (they kept Slack), but as mentioned, their Anthropic costs are now in the millions.
The firm is using AI effectively in various use cases that’s no doubt contribute extensively to this—one example given is Gwen, an AI agent they use to negotiate contracts with healthcare providers—with the CEO noting that the economics still work even if Anthropic increases its prices by five-fold.
Maintenance, as mentioned, remains a sneaky area (with many examples only projecting these costs, or else still in the honeymoon phase with custom solutions). The company itself also becomes responsible for all regulatory mapping, security issues, defects, API adjustments, integration wrinkles, and training needs.
A Retool survey of 817 companies provided some interesting results both directions. 35% of firms had already replaced at least one SaaS function with their own tool, and 51% had built production software already in use with AI.
But only 8% used AI-generated code without changing it manually, and, regardless of tool, just 31% reported prompting their way to the final product.
In other words, most developed pieces of applications rather than working final versions, with just 14% of the builders indicating they were “very confident” in the AI code. By contrast, 20% were not confident, and 6% not confident at all.
What’s clear is that AI is enabling companies with the technical know-how and right size to rapidly accelerate software creation. And for firms with this talent and minimal need for (or who aren’t seeing sufficient return from) their SaaS solutions, this move can be a real cost-saver.
Salesforce ROI, meanwhile, remains highest for businesses with more complex, integrated, or regulated customer operations.
The Salesforce AI integration, industry leaders, and use by AI companies
Of course, it’s interesting to note that Salesforce is still used heavily at many of the top AI companies.
Salesforce CEO Marc Benioff regularly points out that Anthropic, who has also partnered with them, continues to increase its use of their services. (OpenAI, too, is currently listing roles for a Salesforce Ads System Engineer, and data architects to manage Salesforce integrations.)
So while Claude may be helping several 50-person companies to replace Salesforce with AI, Salesforce is at the same time helping Claude’s parent, Anthropic, to maintain, market, and sell Claude to all its customers.
Salesforce is among SaaS firms that are shaking up their seat-based pricing, with many moving increasingly to outcome-based models. And while details are still shifting here, this move is expected to take into account Salesforce Agentforce’s Agentic Work Unit, which tracks completed tasks.
But how is Salesforce Agentforce changing the future of AI-powered CRM?
This has been one of the company’s loudest messages of late, and it’s part of an ongoing effort to solidify their position as a trusted data layer.
While employees may interact less with the Salesforce CRM UI, the core of what it provides still works through AI models (including Claude, and their popular Slack integration, Claude Tag). Underneath, Salesforce is managing customer records, identity and permissions, workflows, and approved actions.
And Salesforce Agentforce adoption has been growing fast. In the company’s May 2026 numbers, annual recurring revenue was up 205% year-over-year, with Agentforce and Data 360’s combined recurring revenue reaching near $3.4 billion. Salesforce also reported processing nearly one trillion API calls across its core offerings in their first quarter alone.
And as a force in enterprise AI automation, Slack’s MCP passed one million users just six weeks after launch.
Is Salesforce still worth investing in for enterprises in the age of AI?
The bottom line for me is this: SaaS providers have to change—and change fast—like many businesses, if they’re going to retain their value to customers.
My title stat is from Gartner, released in July. As their research notes, “This is less an apocalypse and more of a metamorphosis.”
(Salesforce CEO Benioff calls this transformation the “SaaSquatch.”)
This means shifting the focus to outcomes. And companies with long track records and access to customer experiences are better positioned for this than any.
So what companies are successfully adopting Salesforce Agentforce and AI automation?
FedEx is one of many examples touted by Salesforce. The global shipping powerhouse reports increasing their customer activation rate from 25% to 40% in just five months, delivering more than 2,000% ROI by activating dormant customers.
A Salesforce Forrester impact study of global customers provides more insight. While it headlines with striking ROI numbers (396%), the cost details are informative.
The report points to essential costs from AI agents, Salesforce implementation services, training, and ongoing management.
It also showcases 35% customer service case deflection with a 50% reduction in time, and with an important reduction in siloed systems.
What are the biggest challenges businesses face when implementing Salesforce AI solutions?
As mentioned, my company PTP is an established Salesforce consulting partner. This gives us unique visibility on the importance of implementation. We’ve helped companies start from scratch as well as overhaul their Salesforce solutions to help them get the value they were not realizing.
I’m also a firm believer in what AI can do to help businesses transform and scale.
Our AI VOICE framework came into being because of our repeated experiences with companies struggling with AI governance, data quality, workflow redesign, and in general to get real business results without adding headaches and piling up unexpected costs.
For this reason, I remain somewhat skeptical about the lasting power of some Salesforce-replacement stories, at least for companies at enterprise scale.
To answer my opening question, yes, I believe that Salesforce is still worth it.
Provided you’re using it to actually connect your data, people, and AI in ways that are really aligned to your business goals.
References
How Small Firms Use Claude to Quit Salesforce, The Information
Curative CEO says company ditched a $600k-a-year Salesforce contract after vibe-coding a CRM in 2 months, Business Insider
The build vs. buy shift: how vibe coding and shadow IT have reshaped enterprise software, Cache, Retool
Gartner Says $234 Billion in Enterprise Application Software Spend Is at Risk from Agentic AI, Gartner
