Salesforce Insights in 5 Minutes·Issue #196·For architects
Is Salesforce AI Pricing Encouraging Bad Architecture?
Salesforce is pushing structured, secure ways for AI agents (like Claude or ChatGPT) to tap into its platform, using things like AIforce, MCP, and the Headless Toolkit. Makes sense - safer and easier to manage. But here's the catch: Salesforce also plans to charge Flex Credits whenever those AI agents actually use this proper, governed access via MCP or direct APIs. So companies end up with a financial reason to avoid the very method Salesforce says is the safest and cleanest. That could push people toward messier workarounds instead - meaning the pricing model might actually work against good AI architecture.
The site uses the cookies it needs to work, such as your sign-in. With your consent it also uses
Google Analytics, to count visits, and Google AdSense, whose ads fund the
extension: both set their own cookies and send your address to Google. Neither runs until you choose, and you
can change your mind at any time from Cookie settings at the foot of every page.
Which cookies, and why.
