News

Salesforce puts every role on a monthly AI token budget

Salesforce now gives every role a monthly AI token budget, from $100 for reps to $1,000 for technical sellers. What small sales teams can copy.

20 Nov 20265 min readSalesformics editorial
Laptop on a sunlit desk next to a glass jar of coins and a short stack of wooden budget blocks
In short

Salesforce now gives every role a monthly budget for AI tokens. An internal memo from COO Miguel Milano, first reported by Business Insider on 17 November 2026, sets $1,000 a month for technical sales roles and $100 for non-technical roles such as sales reps. The change came after some employees spent thousands of dollars a month on tokens last quarter.

  • Budgets differ by role and department, not one number for everyone
  • A pilot team ran well under budget by month two with no drop in output
  • The memo landed the same week as a new round of Salesforce job cuts

Business Insider reported the token budgets alongside a fresh round of layoffs, citing an internal memo and messages on an internal Slack channel. The Information separately reported on the same memo and its caps on employee AI spending.

The story matters well beyond Salesforce. The company that sells Agentforce to the rest of the market found that unmanaged AI use got expensive inside its own sales and services teams. If that happens at a company with Salesforce's buying power, a 15-person sales team paying list prices for frontier models will feel it sooner.

Salesforce chose a budget per role. The other lever is cheaper work per task. Routine jobs like call summaries, follow-up drafts and CRM field updates rarely need the largest model, and a smaller model trained on your own notes and templates can do them for a fraction of the cost. Teams without machine learning engineers usually buy that work as LLM fine tuning services and keep the frontier model for the hard questions.

What did Salesforce change?

Every role at Salesforce now has a monthly token budget. In his email to staff, Milano said the goal was to focus AI spending on the work that matters most for customers and the business. A Salesforce spokesperson added a caveat: there is no single personal budget for every employee, and team leads manage how their people use AI.

Salesforce tested the approach on one team before the rollout. According to the memo, that team was operating well under budget by its second month, and its output did not drop.

How big are the budgets?

The reported monthly limits vary by department and by how technical the role is:

  • Sales, technical roles: $1,000 a month.
  • Sales, non-technical roles such as sales reps: $100 a month.
  • Customer success and professional services, technical roles: $500 a month.
  • Customer success and professional services, non-technical roles: $100 a month.

That is a ten-to-one gap inside the same sales organization. Salesforce is betting that people who build demos, write code or configure systems get more out of heavy model use than people who mostly draft emails and prep calls.

Why now?

The memo arrived in a hard week for the company. Business Insider reported that layoffs hit staff across several teams and levels, and that the scope of the cuts was unclear. Trade outlets count it as Salesforce's fifth round of job cuts in 2026. The stock fell 9% on Tuesday, 17 November, and Yahoo Finance noted it was heading for its worst week since April.

Cost control on AI and headcount cuts point the same way. Salesforce is under pressure to show that AI makes the business cheaper to run, and an open-ended token bill works against that story.

What does it mean for small and mid-size teams?

  • AI spend needs an owner. Seat licences are easy to count. Token usage is not, and it grows quietly as reps paste whole threads into a chat window.
  • One budget for everyone is the wrong shape. Salesforce split by role for a reason. A sales engineer and a junior SDR use AI in very different ways.
  • Usage-based pricing is spreading. Vendors already bill AI agents per action or per lead, as we covered with Microsoft's sales development agent. A per-role budget is the internal mirror of that pricing.
  • A cap can coexist with output. The pilot result suggests people trim waste, not useful work, once they see a number.

How do you set AI budgets for your own team?

  1. Measure a month first. Pull usage by user from your AI vendor's admin console before you set any limit.
  2. Group roles by how they use AI. Builders and analysts in one tier, reps and coordinators in another. Two or three tiers are enough for a small team.
  3. Move routine tasks to cheaper models. Summaries, tagging and first drafts are the easiest wins. Keep the large model for research and complex deals.
  4. Review after two months. Compare spend with pipeline activity, not with last month's spend. Raise the limit where AI clearly shortens the sales cycle.
TipPut AI spend next to the tools it supports when you review your sales stack. Our sales enablement tools guide shows how to judge each tool by the time it saves reps.
Try it on your own pipeline14-day trial, no card needed.
Start free