Agentforce Is Changing How You Pay for Salesforce. Get Ready Before Your Next Renewal.

For twenty years, budgeting for Salesforce was simple: count the users, multiply by the license price, add the annual uplift. That model is going away.

Salesforce now sells AI agents in three ways at the same time: per-user add-ons, flat-fee editions and pay-as-you-go credits. This month it renamed its top AI tier "Max Edition," and on October 20 it agreed to buy Braze for about $5.66 billion to push further into AI-driven marketing. Other CRM vendors are moving the same way. Gainsight just announced it will move away from per-seat pricing in 2027.

I've sat on both sides of enterprise software negotiations. I can tell you that when a vendor changes its pricing model, the customer who walks in unprepared pays for it. If your Salesforce renewal falls in the next six to twelve months, here's what I'd put in front of your leadership team.

What Actually Changed

• •Three ways to pay for AI. Agentforce is offered as per-user add-ons (around $125/user/month), all-in editions (Max at roughly $550/user/month) and consumption via Flex Credits ($500 per 100,000 credits, or about $2 per customer conversation).
• •Seats are no longer the only meter. With credits, your bill depends on how often agents run, not how many people log in. That's hard to forecast when you've never run an agent.
• •The platform keeps growing. Data Cloud, Agentforce and now Braze are all being sold as parts of one "agentic" platform. Every new piece is another line item someone will try to bundle into your renewal.
• •Standard uplifts haven't gone away. Licensing advisors still report annual renewal increases in the 8–10% range unless they are negotiated down.

The Real Risk Isn't the Price. It's Buying Before You're Ready

Gartner expects more than 40% of agentic AI projects to be canceled by 2027, mostly because the ROI was unclear or governance was weak. In my experience, the CRM version of that failure looks like this:

• •Agents are switched on over messy, duplicate customer data.
• •The sales or service process was never documented, so the agent automates confusion.
• •Nobody owns the consumption budget, so the first true-up invoice is a surprise.
• •Credits are bought "use it or lose it" based on a vendor's estimate, not your volume.

An AI agent is only as good as the process and data underneath it. That hasn't changed. What's changed is that the vendor now gets paid whether the agent helps you or not.

How to Size Agentforce for Your Business

Before you commit, get these answers in writing:

• •Pick one or two use cases with clear volume: service case deflection, lead qualification or order status. Not "AI everywhere."
• •Count your monthly transactions for those use cases today: cases, chats, inbound leads.
• •Model three scenarios (low, expected, high adoption) on both seat-based and credit-based pricing. Note where they cross over.
• •Check your data first. Duplicate accounts, missing fields and disconnected ERP data will drive up both cost and errors.
• •Name an owner for AI consumption the same way someone owns your cloud bill.

What to Negotiate at Renewal

• •A cap on annual uplift for your core licenses, separate from any AI purchase.
• •A small Flex Credit pilot pool instead of a multi-year AI commitment.
• •The right to move between add-on, edition and consumption models as your usage becomes clear.
• •Usage reporting you can see monthly, not just at true-up.
• •No forced bundling of Data Cloud or new acquisitions you haven't evaluated.

What to Do Next

• ☐Find your Salesforce renewal date and when the quote is due.
• ☐Pull a list of current licenses, and how many are actually used.
• ☐Pick one high-volume use case and measure its monthly transaction count.
• ☐Run a quick data-quality check on accounts, contacts and cases.
• ☐Build a three-scenario cost model before your account executive builds one for you.
• ☐Decide who owns AI consumption spend going forward.

If you're also planning 2027 budgets, see my earlier piece on IT budgets built on prices that no longer exist. And if your processes aren't ready, read why operational transformation comes before AI.

Frequently Asked Questions

Related Reading