> ## Documentation Index > Fetch the complete documentation index at: https://docs.guild.ai/llms.txt > Use this file to discover all available pages before exploring further. # Optimizations > Find a cheaper configuration for an agent already running in your workspace, without giving up quality. **Optimizations** looks at an agent that's already running in a workspace and tests cheaper ways to run it — a smaller model, a trimmed tool list, a tighter prompt — against the agent's own recent production traffic. Nothing changes until you review the results and accept a change. Where Insights and Spend Explorer show you what an agent costs, Optimizations is where you act on it. Optimizations is currently being offered to a select group of customers. If **Optimizations** doesn't appear in your sidebar, the feature isn't enabled for your account yet. ## Starting a run Open **Optimizations** in the sidebar, or click **Optimize** next to a workspace in the [Workspaces breakdown](/insights/usage#workspaces) on the Insights page. A workspace can only be optimized if it's eligible: * One agent in the workspace — the **entrypoint** — must be owned by your account, still installed in the workspace, and have at least 10 recent sessions. Guild picks the workspace agent with the most recent activity as the entrypoint automatically. * At least one agent in the workspace (the entrypoint or a sub-agent it calls) must actually appear in those sessions, so there's something to measure. * The workspace can't already have a run in progress. Only one optimization run is active per workspace at a time. A run defaults to a \$50 spend budget, split across every agent it measures, and stops once that budget is spent. ## How a run works ### Building a baseline evaluation Optimizations replays an agent's own history to judge whether a cheaper configuration is still good enough, so the first step is turning that history into an evaluation. Guild takes the entrypoint's most recent sessions (10 by default) and reduces each one to what actually happened — the triggering input, the tool calls made, and the final outcome — rather than the raw conversation. An LLM turns each of those digests into an evaluation sample: the original trigger becomes the input, and the outcome becomes what a judge checks the optimized agent against. Before that evaluation is used to compare anything, Guild runs the agent's current, unmodified configuration against it. Since every sample is drawn from a session the agent already completed successfully, the current configuration is expected to pass. If it doesn't, Guild diagnoses why (a missing fact, a reference that's too strict, or a sample that can't be reproduced) and repairs the evaluation, for up to two rounds, before measuring any candidates against it. You can view the generated evaluation from the run screen. Editing its samples isn't available from the Optimizations UI today. ### Testing cheaper configurations Once the baseline evaluation holds up, Guild tests configuration changes — called candidates — against it: | Candidate | What it changes | | ------------------ | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | **Model** | Swaps the agent's model for a cheaper one on the same provider's model family and re-runs the evaluation. | | **Tool trim** | Compares the agent's declared tools against the tools it actually called across the sampled sessions, then re-runs the evaluation with the unused ones removed. | | **Prompt rewrite** | Has an LLM propose a tighter version of the agent's system prompt, informed by the prompt itself and the sampled sessions, then evaluates the rewrite. Only available for agents with a static system prompt. | | **Combined** | Once a model swap and a tool trim have each been evaluated on their own, the winning model and the winning tool trim are combined into one candidate and evaluated together. A prompt rewrite is never combined with a model swap, since a prompt tuned for one model may not transfer to another. | Each candidate is actually built and run against the same evaluation as the baseline — Guild doesn't estimate savings, it measures them. A candidate only counts as a real result once enough of its trials complete and enough samples score on both the baseline and the candidate to compare fairly. ### Scoring the result Every candidate gets a verdict based on how its quality compares to the baseline: * **Positive** — quality held (within a small tolerance) or improved. * **Neutral** — quality dropped slightly, within a wider tolerance. * **Negative** — quality dropped meaningfully, or the candidate lost tool calls or checks the baseline passed. Cost is priced in dollars at current list prices, not raw token counts, since a cheaper model can sometimes need more tokens (more tool calls, more turns) to finish the same work. Only a candidate with a positive verdict that's also cheaper than the baseline is ever recommended. ## Reviewing and accepting changes The run screen lists every agent Guild measured, with its candidates side by side — baseline, model only, tool trim, prompt rewrite, and combined — showing the quality change and cost change for each. A candidate that lost a tool call or check the baseline passed is marked **Behavior changed** regardless of its cost. For each agent with a recommended change, you can: * **Accept** — installs the recommended configuration for that agent. Guild publishes it as a new agent version, points the workspace at that version, and turns off autoupdate for the agent so a later publish doesn't silently overwrite the optimization. This can take a few minutes and can't be stopped once it starts. * **Accept all** — accepts every agent in the run with a recommended change in one step. * **Dismiss** — declines the recommended change for that agent. This can't be undone from the run screen. * **Cancel** — stops the whole run. Any agent still awaiting a decision is dismissed. There's no one-click rollback for an accepted change today — reverting means publishing another agent version the way you normally would.