AI budgets affect how people work. A developer who regularly reaches their limit may need more room to keep shipping. Someone else may use only a small part of the same allowance. As teams use more AI tools, managers need to understand both the cost and the work that comes with it.
Today we’re introducing AI Budgets, a new feature in the Antenna control plane. It brings AI spend, forecasts, individual limits, and delivery data together so you can decide where to allocate budget across the entire company.
AI Budgets helps you answer the questions behind those decisions:
- Who is spending the most on AI, and on which tools?
- Who is on pace to exceed their spending limit this month?
- Who could benefit from more AI budget, based on their output and usage?
- Where have we allocated more budget than people are likely to use?
AI spend tracking and forecasts
AI spend is harder to predict than most software costs. Usage-based pricing rises and falls with how each person works, and the charges are often split across vendors that bill separately, so the full picture may only emerge once the invoices arrive. A recent McKinsey survey found that few companies have mature practices for tracking or forecasting this spend.
AI FinOps maturity is generally low
Share of surveyed companies reporting mature practices for managing AI costs.
- Spend visibility and attribution
- 25% of surveyed companies reported mature capabilities.
- Token consumption tracking
- 24% of surveyed companies reported mature capabilities.
- Cost allocation (chargeback/showback)
- 21% of surveyed companies reported mature capabilities.
- Forecasting and budgeting
- 20% of surveyed companies reported mature capabilities.
McKinsey & Company, The Cost of Intelligence (July 2026).
Antenna's AI Budgets centralize costs from your connected AI tools, including seat fees and variable usage. You can track spending against each tool’s budget, see which tools account for a person’s costs, and identify who is on pace to exceed their limit.
Month-end forecasts give you time to govern and control usage in real-time. If a tool is heading over budget, you can examine the people and usage behind that forecast before deciding whether to adjust limits or increase the budget. Past months provide a record of actual spend, and an annual view for each tool supports planning across your fiscal year.
This gives managers a shared basis for discussing AI costs across vendors, teams, and departments, with enough detail to act on an individual’s needs.
AI budget allocation that adapts to usage
Spending more on AI can be a strategic decision when it supports higher productivity. AI Budgets lets you compare how budget and spend are distributed with how work is delivered.
You can compare groups of developers by their share of deliveries, AI spend, and allocated limits. You can also examine each person’s output alongside their spend and their team’s median. New Deliveries, Accepted PRs, and Engineering Hours offer different ways to assess that work.
For example, a developer who delivers consistently and is forecast to reach their limit may be a good candidate for additional budget. A person who uses only a small part of their allowance may need a lower limit. These comparisons give you evidence for reviewing allocations as people’s needs change.
Antenna recommends increases for people with high throughput who are forecast to exceed their limits, and reductions for people forecast to use substantially less their than limits. You can apply recommendations individually or to several people at once. Additional grants last until month end, so you can support a busy month without committing to a permanent increase.
Right-sized limits lead to more efficient use of AI. Industry survey data shows that when developers know how much budget they have, they make more deliberate choices about how to spend it. Seen this way, a budget is less a limit on productivity and more a guide people can plan their work around.
Usage limits shape how developers use AI
Share of developers reporting changes in how they personally use AI tools at work due to costs or usage limits. Selected responses from 1,033 participants; multiple answers allowed.
- Monitor their own AI usage
- 32.6% of survey participants selected this answer.
- Choose cheaper models
- 25.3% of survey participants selected this answer.
- AI usage is metered to control costs
- 23.8% of survey participants selected this answer.
- Become more selective about tasks
- 16.6% of survey participants selected this answer.
- Team switched to cheaper models or vendors
- 6.9% of survey participants selected this answer.
Stack Exchange Inc. (2026). “AI cost and usage limits”. Stack Overflow Developer Survey 2026. Licensed under ODbL 1.0. https://survey.stackoverflow.co/2026/knowledge/data/ai-cost
Predictive spend alerts
Alerts help you keep track of changing usage without checking every person’s spend each day. You can notify a person, their manager, and tool admins when spending reaches a chosen share of that person’s limit.
Daily warnings flag people who are over their limits or on pace to exceed them, tools that are over budget, and unusual spending patterns. A weekly digest brings together the month’s position against your budget and the people who may need adjustments.
Together, forecasts and alerts give you time to investigate changing usage patterns and revise budgets while those decisions can still affect the month.
Automated spend limits
For teams managing many individual limits, automatic spend limits can handle routine adjustments. You can group developers into groups based on output, set a monthly limit for each group, and define approval policies for additional budget.
Before enabling automatic budgets, you can review the effect on total limits and forecast spend. Antenna then reviews limits each day, while preserving any limits set by hand. Automatic increases last until the end of the month.
You can also sync individual limits back to your AI tools for enforcement. At launch, we support syncing limits with Claude Enterprise and ChatGPT, but we plan to add more providers in the future.
Spend and limits on other connected tools remains visible for monitoring and planning.
Getting started
Getting started takes only a few minutes. Connect your AI tools, review their historical costs, and set tool budgets and individual limits. Use forecasts and alerts to follow spending through the month, then review delivery data to decide where additional budget would be useful.
Learn more about AI Budgets, or request a demo to discuss using it with your team.