Be the answer for AI
Use Gauge Chat to improve your brand’s AI visibility, or Gauge Agents to understand how coding agents choose and use your tool.
Optimize for Chat experiences
Use Gauge Chat to measure your brand’s visibility in AI answers, improve and create content based on data, and measure again to see what changed.
View Chat Docs →Optimize for coding agents
Use Gauge Agents to measure whether coding agents choose your product and can use it successfully, then test improvements.
View Agent Docs →Popular starting points
Gauge Chat quickstart
Gauge Agents quickstart
Build with Gauge
Work with both products from your terminal, your AI client, or your own code. One organization API key works everywhere.
Gauge CLI
Gauge MCP
Gauge Chat API
Gauge Agents API
What Gauge does
Gauge helps companies understand and improve how AI systems represent, recommend, and use their products. Traditional analytics show what people click after they reach a website. Gauge measures the decisions AI makes before and during that journey: which brands appear in an answer, which sources shape the answer, which developer tools a coding agent recommends, and whether the agent can use those tools successfully.
Gauge turns that evidence into action. Teams can trace a result back to the prompts, answers, citations, agent traces, and code changes behind it; identify the gaps causing lost visibility or failed adoption; and test changes to content, documentation, positioning, and product experience. The goal is not simply to monitor AI. The goal is to help your product become the answer AI gives and the tool agents choose.
Gauge’s two products
Gauge Chat
Gauge Chat is the AI visibility product for marketing, growth, and content teams that want their brand to appear accurately and prominently in AI search. It runs customer questions across major AI platforms every day, measures mentions and citations, compares competitors, and shows the sources behind every result. Ask Gauge and the Content Engine turn visibility gaps into recommended actions and publish-ready content.
Use it to: monitor AI answers, understand competitive visibility, find content and positioning gaps, and improve the information AI uses to describe your brand.
Gauge Agents
Gauge Agents is the Coding Agent optimization product for developer-tool companies that want coding agents to choose their product and implement it correctly. It runs real coding agents in real repositories, records their research and code changes, and measures both Agent Preference—which tools agents recommend and adopt—and Agent Experience—whether agents complete a task successfully. Teams can use the full web product or the Gauge CLI; both provide the same measurement, run, and results workflows.
Use it to: compare agent and model behavior, find documentation or onboarding friction, understand why competitors win, and test which changes improve selection and implementation.
How to get started
Start with Gauge Chat
- Sign up for the Growth plan.
- Add your brand, website, competitors, and the customer questions you want Gauge to track.
- Review the daily answers, citations, visibility metrics, and recommended actions.
Start with Gauge Agents
- Choose local runs in Docker with your own agent credentials, or hosted runs with automatic judging and shared results.
- For the web product, sign in at agents.withgauge.com and complete the guided onboarding flow.
- For the CLI, install
@withgauge/cli. Usegauge init —localto start locally without a Gauge account, orgauge onboardandgauge config set backend hostedfor hosted runs. - Create an Agent Preference or Agent Experience measurement, configure its runs, and inspect the results.
Choose the outcome you want to improve
Both Gauge products connect measurement to action. Choose the one that matches how customers—or their coding agents—discover and adopt your product.
Understand and improve how AI represents your brand.
See where your brand appears, which sources shape the answer, and what your team can do next. Gauge Chat combines daily measurement with analysis and publish-ready content.
Measure why coding agents choose your product—and whether they succeed with it.
Run real agents in real repositories. See their research, decisions, code changes, and failures so your team can improve documentation, onboarding, and product experience.
Each run uses 5 infrastructure credits plus AI token credits. Actual usage varies by model and task. See credit details.
What you can do with Gauge
Follow the evidence from an AI answer or coding-agent session to a specific improvement. Here are the core workflows in each product.
Gauge Chat: find where competitors win in AI answers
Track the questions buyers ask and inspect brand mentions and rankings, sentiment, competitor visibility, and cited pages. Gauge tracks ChatGPT, Perplexity, Gemini, Google AI Overviews, Google AI Mode, Microsoft Copilot, Claude, and Grok. Coverage depends on your enabled engines and plan; Claude and Grok are available through a bring-your-own-key add-on.
Ask Gauge “Which of our pages get cited without our brand being mentioned?” or “Where did competitors appear more often than us on ChatGPT?” Open the supporting answers and sources to decide what to change. A citation shows that AI used your page; a mention shows that it named your brand. Explore Ask Gauge.
Gauge Chat: turn a content gap into a published improvement
Create a researched article, comparison, or alternatives page grounded in your visibility data. Review the brief and outline, then edit the draft using saved company facts and style guidance. Update existing pages or publish through Webflow, Framer, Sanity, or a GitHub pull request; export Markdown for other platforms. Track changes in citations, mentions, and visibility for the same prompts and engines after publishing. Follow the publishing workflow.
Connect Google Analytics and Search Console, and PostHog to bring traffic, search, and product data into your analysis. Save repeatable workflows as Skills, schedule competitor reviews, and send reports to Slack or email. Gauge Chat complements SEO and analytics tools with the answers and sources that shape AI discovery.
Gauge Agents: measure product selection and implementation separately
Gauge supports Claude Code, Codex CLI, OpenCode, Pi, and Antigravity. Run realistic tasks in isolated workspaces with your repositories, docs, Skills, packages, and MCP servers. Inspect research, tool calls, final responses, and code diffs. Use gauge models list to see available agent/model targets for your selected backend.
Agent Preference: ask “Choose an authentication provider and add sign-in to this Next.js app.” Gauge records ranked picks, the top recommendation, actual adoption in code, and the evidence behind the choice. Unbranded tasks measure unaided preference; head-to-head comparisons explain tradeoffs between named products. Measure product selection.
Agent Experience: name your product and define observable success criteria, such as sending an email through your API, reading the key from an environment variable, and keeping secrets out of source code. Gauge’s hosted LLM judge checks each criterion against session evidence. Compare pass rates and recurring failures across agents and models to find onboarding, documentation, or API friction. Measure implementation success.
Gauge Agents: test changes before you ship
Commit tasks and criteria alongside your code. The Gauge GitHub App makes your candidate package, CLI binary, docs preview, Skills, or files available to coding agents, then reports results in a PR comment and status check. See whether agents can use the changed product before merging, or run cases from your own CI with the CLI. Set up PR evaluations.
To improve a docs page, Skill, or CLI, start from an issue agents hit. Gauge edits copies of the files the agents used, reruns those sessions with the edits, and shows which findings resolved. Review the diff and publish it through your normal process. Improve agent experience or improve agent preference.
Gauge Agents: start with the pages agents already visit
Run a public website assessment with AX Check, or connect Cloudflare or Vercel access logs to Agent Traffic to see which pages recognized coding agents request. Turn an important finding into a repeatable eval. Page requests show what agents fetch; preference and experience tests reveal what they choose and whether they succeed.
Run the core improvement loop locally for free. Test real coding agents against your product, inspect where they struggle, change your docs or Skills, and rerun the task to see whether the fix worked—all without a Gauge account or Gauge credits. The CLI runs sessions in Docker, and your coding agent judges the results and publishes findings. Bring your existing agent subscription or API credentials; your provider’s usage charges still apply. Start testing locally for free.
Upgrade to managed Gauge for more capacity, configuration options, and sharing with your team. Get automatic judging, shared results, scheduled runs, GitHub PR checks, and optimizations. Hosted sessions use Gauge credits. Get started with managed Gauge.