Salesforce plans to charge customers when outside AI agents access its platform. Here's why that matters, and why we believe businesses should always have the freedom to choose.
Imagine building an AI agent that automatically updates customer records, analyzes sales opportunities, and generates reports using data your company already has in Salesforce. You built the agent. You're paying for the AI model. You're already paying Salesforce for the software. Now, Salesforce wants to charge you when that agent interacts with its platform.
That's the direction Salesforce is heading, according to recent reporting from The Information and the company's own documentation.
And it raises a pretty important question: Why should choosing an outside AI provider mean paying more to use software your business already pays for?
What Salesforce Is Changing
Salesforce is preparing a new billing model for external AI agents, including agents built using platforms outside its own ecosystem.
According to Salesforce's documentation, the changes include:
Agent registration: Customers will need to register AI agents that access Salesforce, giving each agent its own identity and permissions.
Usage tracking: Salesforce will track successful calls made by registered agents through APIs or the Model Context Protocol (MCP).
New charges: Those interactions will consume Salesforce Flex Credits, creating an additional usage-based cost.
Salesforce is targeting November 2026 for the rollout, although the actual pricing hasn't been announced. The company says it will provide 30 days' notice before metering begins. Existing customers using agents through APIs will generally transition at contract renewal. Traditional, non-agentic integrations aren't affected by this new billing model.
To be fair, there are legitimate reasons to register and monitor AI agents. Businesses need security, access controls, and visibility into what automated systems are doing.
But there's a difference between governing access to a platform and charging customers every time their own AI uses it. And that's where things get interesting.
The Bigger Problem: Who Controls Your AI?
For years, enterprise software companies built their businesses around subscriptions. You paid for employees to access applications, and those applications became central to how your company operated.
AI agents are changing that relationship.
Instead of having employees log into five different systems to complete a task, businesses can build agents that work across those systems automatically. Your AI agent might pull information from Salesforce, analyze it using an open-weight model, update another application, and deliver a report without anyone opening a traditional software interface.
That's a pretty significant shift. It also creates a problem for software companies whose revenue depends on customers using their applications directly.
As The Information reported in October 2026, some Salesforce customers are already reconsidering their AI strategies and potentially moving data elsewhere if the new fees become too expensive. The concern isn't just what Salesforce might charge. It's what happens if every major software vendor adopts a similar approach.
Imagine paying for your AI model, your agent infrastructure, your software subscriptions, and then an additional fee every time your agent interacts with one of those applications. Suddenly, the automation you built to save money starts accumulating charges across your technology stack. And the more successful your AI becomes, the more expensive it could be to operate.
AI Should Give Businesses More Freedom, Not Less
We think the future of enterprise AI should be built around choice. Businesses should be able to choose which AI models they use, where they run them, what tools they connect, and how their data is managed. They shouldn't have to commit to one model, one cloud, or one company's AI ecosystem just because that's where their data happens to live.
This is particularly important because AI is evolving so quickly.
The best model for a task today might not be the best model six months from now. A cheaper infrastructure option might emerge. A new agent framework might outperform the one you're currently using.
Businesses need the flexibility to adapt without rebuilding everything or getting trapped by their existing technology decisions. That doesn't mean software companies shouldn't charge for the value they provide. Of course they should.
But we believe customers should stay with a platform because it delivers value, not because leaving is too difficult or expensive.
At Rival, We Don't Believe in Lock-In
This is something we've been thinking about as we build Rival. Our goal is simple: Give businesses one place to build, run, and manage AI without forcing them into a single model, cloud, or technology provider. And that philosophy extends to something even more fundamental.
Your data should be yours. Period.
That's why we've introduced a new data export feature in Rival.
You can now download your data directly from the platform without submitting a support ticket, negotiating with anyone, or jumping through hoops.
Here's how it works:
Individual users can export the data Rival holds about them from My Profile in JSON or CSV format.
Organization Owners and Admins can export their organization's data directly from the Command Center.
Exports are designed for portability, download your data in JSON or CSV format, with sensitive connector credentials excluded for security.
No complicated process. Just an export button. We believe businesses should have the freedom to take their data elsewhere, whether they're changing platforms, building new integrations, or simply maintaining their own records. And while data portability alone doesn't eliminate every kind of vendor lock-in, it's an important place to start.
The Future of AI Should Be Open
Salesforce's proposed pricing model is a reminder that the economics of enterprise AI are still being worked out. As agents become more capable, software companies will have to rethink how they deliver value and how they charge for it.
We understand that. But we also believe businesses should have more control over their technology, not less. More choice in models. More flexibility in infrastructure. More visibility into costs. And more ownership of their data.
At Rival, we'd rather earn your business every day than make it difficult for you to leave.
That's the kind of AI ecosystem we're building.
Sources
Salesforce Help: Understand AIforce Impact: September 17, 2026. Official details on agent registration, metering, and planned rollout.
The Information: October 2026 reporting on Salesforce's external AI agent pricing and customer reactions.
Salesforce MCP Pricing Explained: Headless Platform Interactions and Flex Credits: October 23, 2026. Explanation of how Salesforce's proposed usage-based billing would work.
Salesforce: Data Cloud Pricing Updates: September 12, 2025. Background on Salesforce's consumption-based pricing strategy.