Guild for Cost Control: Cut AI Spend, Prove Agent ROI
Know what agents cost. Prove what they deliver.
Agents are expensive. But they don’t have to be. Monitor spend, reduce token costs, and deliver more ROI from every agent with our Insights Dashboard.
Spend Breakdown
This month
| Agent | Cost |
|---|---|
| PR Reviewer | $312.40 |
| Issue Triage | $186.20 |
| Standup Summary | $142.80 |
| Deploy Notifier | $118.50 |
| Doc Sync | $82.40 |
| Release Notes | $64.10 |
Usage Cap 82% of cap
PR Revieweragt_8c3d51e09a Monthly
Spent this month $0.00/ $380 cap
Alert threshold 80% $380 cap
Alert at 80% · $304 Triggered
Hard stop at 100% · $380 Armed
Enforced automatically Resets Sep 1
The right tools to manage token costs at scale
Minimize your token spend with Insights Dashboard: the first centralized tool to understand how agents consume resources, drive usage, and incur costs—and then give you the resources to manage it in real time.
Cost by PR
pr-reviewer #4821
Last 7 days $-83.92
−18% vs last week ↓
| PR Number | Description | Cost |
|---|---|---|
| #2437 | Fix flaky auth test | $0.68 |
| #2418 | Refactor session store | $2.14 |
| #2402 | Add retry to webhook handler | $3.41 |
| #2396 | Bump runtime deps | $0.42 |
| #2381 | Split billing migrations | $2.87 |
5 most recent runs 38 runs this week
Run cost pr-reviewer #4821
Total $-31.79
| PR | Duration | Plan | Cost |
|---|---|---|---|
| #2418 | 4m49s | claude-sonnet-418.4k→2.1k | $0.09 |
| toolgithub_pulls_get | $0.00 | ||
| toolgithub_pr_diff | $0.00 | ||
| 132.5k | subagentsys-coding | $0.61 | |
| reviewclaude-opus-488.2k → 6.4k | $1.32 | ||
| toolslack_post_msg | $0.12 | ||
| 978k cached reads | saved $0.86 |
AI observability
Follow the money
Visualize token usage by workspace, agent, user, provider, and model to see where your budget is going—and where to improve.
Spend Limits
Org-wide
Monthly Cap
| Team | Budget | Usage |
|---|---|---|
| Platform Team | $800 | $0 |
| PR Reviewer | $150 | $0 Near limit |
| Issue Triage | $180 | $0 Over limit |
| Deploy Notifier | $100 | $0 |
5 limits configured 1 over · 1 near limit
Moderation & publishing
Deploy evals to reduce costs
Create evals that go beyond measuring outputs to programmatically improve token usage.
Evals 12 runs
Determinism score -835% +4pts vs last week
| Feature | Success Rate |
|---|---|
| Retrieves order status | 12/12 |
| Formats currency consistently | 12/12 |
| Escalates refund request to human | 9/12 |
| Cites source doc in response | 11/12 |
| Declines out-of-scope request | 12/12 |
4 of 5 cases stable Last run 2m ago
Budget Limits
Control spend automatically with rate limits
Take control on day one. Identify costly agents, set spending caps to reduce token usage, and control ROI out of the box.
Agent Spend This Month
Total $-2408.14
+12% from last month ↑
| Agent | Cost |
|---|---|
| PR Reviewer | $0.00 |
| Issue Triage | $0.00 |
| Standup Summary | $0.00 |
| Deploy Notifier | $0.00 |
| Doc Sync | $0.00 |
Top 5 of 47 agents 12.8M tokens
Real cost
Connect dollars to tokens
Manage your budget better by understanding the financial impact of every agent in production—not just the token counts.
Period over period
This month
| Amount | Last Month |
|---|---|
| $-2479.06 | $752.05 |
+12% from last month ↑
W1 W2 W3 W4
This month Last month 4 wks
Optimize
Track cost control period-over-period
Compare costs across a 7/30/90-day window to evaluate your cost control programs and report on the results.
Integrate all the tools they need—and then some.
Building Guild isn’t just extensible, it’s neutral by design. Connect everything your team runs on—from model to data source. No migrations required.
Models
Guild keeps you flexible, so agents aren’t tied to a single model vendor or setup.
- Anthropic
- OpenAI
- Gemini
- Deepseek
- Cursor
- Codex
Tools
Connect the systems your team already runs on—no migrations required.
- Jira
- Linear
- Slack
- Notion
- New Relic
- GitHub
Frequently asked questions
Why do AI agent costs spiral out of control?
Three reasons: no visibility (teams don't know which agents are driving spend), no attribution (spend rolls up by provider, not by agent or workspace), and no enforcement (alerts fire after the money is already spent). Guild solves all three: per-agent attribution, real-time dashboards, and hard budget limits that actually stop overruns.
How does Guild help me reduce AI agent spend?
Guild identifies your top cost drivers, recommends cheaper model configurations backed by real evals against your production sessions, and lets you enforce monthly budgets with circuit breakers. You cut costs where quality holds and enforce limits where it doesn't.
What are Budget Limits, and how do they stop cost overruns?
Budget Limits are hard spend limits set per workspace, agent, or team. You get notified at 80% of the budget. At 100%, the limit is reached and further usage is automatically blocked until the cycle resets. Alerts warn you. Circuit breakers stop the spend.
Can Guild compare AI spend before and after an optimization?
Yes. Guild tracks spend across 7-, 30-, and 90-day windows so you can measure the impact of an optimization directly: cost per run before, cost per run after, delta in dollars, and quality delta from the eval. Optimization results are measurable, not anecdotal.
Does Guild show real dollar cost, or just token counts?
Both. Token counts by workspace, agent, user, provider, and model, and normalized dollar cost across every provider. Understand the financial impact of every agent in production, not just how many tokens they consumed.
Transform with Guild.
See Guild in action.