# Reflect > Reflect is the guidance system for production AI agents: a developer writes a rule once and defines where it applies, and Reflect delivers it at that step of the run, shows which conversations it reached, and measures whether it helped. It also surfaces the mistakes an agent keeps making, and works with any agent framework and any model. ## Earlier product Reflect does not rebuild the systems your agent calls or run your agent against a replica of them. Older pages, listings and posts may describe an earlier Reflect product: a staging environment, digital twin or replay-testing tool for AI agents. That product is no longer offered, and runreflect.com now redirects to this site. For anything current, use this site and its documentation. ## Product - [Homepage](https://usereflect.dev/): Product overview and illustrative examples. - [How it works](https://usereflect.dev/#how-it-works): Write a rule once, define where it applies, see which conversations it reached, and measure whether it helped. - [Pricing](https://usereflect.dev/pricing): Plans, usage rates and example charges. - [Start free](/signup): Sign up and start on the free Hobby plan, no call needed. ## For coding agents - [agents.md](https://usereflect.dev/agents.md): Step-by-step instructions for adding Reflect to a codebase. Start here if you were asked to integrate Reflect. - [Pricing as Markdown](https://usereflect.dev/pricing.md): Plans, the turn definition and the full comparison. ## Pricing - Hobby: $0 a month. 1,000 turns a month, free. Try Reflect on your agents. Includes: Unlimited agents; SDK, API and MCP access; Find recurring mistakes; Write and preview rules. - Growth: $299 a month, plus $0.005 per turn. For agents running in production. Everything in Hobby, plus: Founder onboarding; Integration help; 24/7 support; Usage dashboard and estimated charges. - Enterprise: Custom. Volume rates on a usage commitment. For high volume and custom terms. Includes: Unlimited agents; Pricing and billing terms in your contract; Support terms agreed in writing; Help planning larger rollouts. - A turn is one run of your agent: everything it does from start to finish, including every model call, tool call and subagent. For most agents that is one user message and the full answer. A run your agent starts on its own, like a scheduled job, is also one turn. A turn counts whether or not a rule applies. If your app runs the agent again, that is a new turn. Examples: a conversation your agent runs 10 times for, once per message is 10 turns; one run that makes 20 model and tool calls is 1 turn; one scheduled run that starts 5 subagents is 1 turn. Never billed: Network retries and duplicate uploads of the same run; Requests Reflect rejects or fails to process; Hobby usage and imported history. - Example Growth charges before credits and tax: 50,000 turns, $549; 100,000 turns, $799; 250,000 turns, $1,549; 1,000,000 turns, $5,299. - Billing: the base fee is billed at the start of each monthly period and usage at the end. Every turn is billed individually, never rounded up to a block. No charge per agent, rule, MCP read or rule lookup. Your agent's own model and tool costs stay with your providers. Growth rates are locked for 6 months from the first payment. Cancellation takes effect at the end of the paid period. ### Pricing questions #### How many agents can I create? Every plan includes unlimited agents, with no agent fees. Add and manage all your agents in one place. Hobby’s 1,000 free monthly turns are shared across all your agents. Growth has one monthly base fee, plus the turns your agents use. #### What is a turn? One run of your agent and everything it does from start to finish: every model call, tool call and subagent. For most agents that is one user message and the full answer. A run your agent starts on its own, like a scheduled job, is also one turn. #### What’s never billed? Retries, duplicate uploads, requests Reflect rejects or fails to process, and Hobby’s 1,000 free turns a month. There’s no charge for agents or rules either. #### When am I billed? The base fee is billed at the start of each monthly period, and that month’s turns at the end, on the same invoice cycle. #### Will my price go up? Not for 6 months. Growth rates are locked for 6 months from your first payment, and they don’t change automatically when that period ends. #### How do I cancel? You can cancel Growth anytime. Cancellation takes effect at the end of the billing period you’ve paid for, so you keep Growth until then. Turns from that final period are billed on the last invoice. ## Documentation - [Documentation](https://usereflect.dev/docs): Product concepts and integration guides. - [SDK quickstart](https://usereflect.dev/docs/quickstart): Install and connect the SDK. - [SDK reference](https://usereflect.dev/docs/sdk/reference): Runtime APIs and configuration. - [Trace capture](https://usereflect.dev/docs/guides/trace-capture): Send agent conversations and tool activity. - [Rules](https://usereflect.dev/docs/guides/corrections): Review, preview, and activate scoped rules. The docs call them corrections. - [Runtime modes](https://usereflect.dev/docs/guides/runtime-modes): Control delivery and validate behavior. - [Troubleshooting](https://usereflect.dev/docs/guides/troubleshooting): Diagnose integration problems. - [Documentation Markdown index](https://usereflect.dev/docs/llms.txt): Links to individual Markdown pages. - [Full documentation text](https://usereflect.dev/docs/llms-full.txt): Combined documentation for text readers. ## Guides - [What is an agent correction? Scoped fixes for AI agents](https://usereflect.dev/resources/glossary/agent-correction): An agent correction is a short instruction an AI agent gets only when its trigger matches. How it works, and how it differs from prompt lines, skills and guardrails. - [Why your AI agent keeps making the same mistake in production](https://usereflect.dev/resources/ai-agent-keeps-making-same-mistake): AI agents repeat mistakes when a fix lands in the wrong place, nobody groups the failures, or nobody measures the change. How to find the pattern and make a fix stick. - [AI agent observability vs evals vs corrections](https://usereflect.dev/resources/vs/observability-vs-evals-vs-corrections): Observability shows what an agent did, evals test a change, and corrections change what it gets in one situation. How they fit with Langfuse, LangSmith and Braintrust. - [Self-improving AI agents: what works in production](https://usereflect.dev/resources/self-improving-agents): What self-improving agents do, where automatic prompt rewriting falls short, and why a review step between a lesson and live behavior is worth keeping. - [System prompt too long? What to do instead of adding more rules](https://usereflect.dev/resources/system-prompt-too-long): How to tell when a system prompt has grown too far, what to move out of it, and how to add an instruction only in the situations where it applies. ## Frequently asked questions ### What does Reflect do? Reflect is the guidance system for production AI agents. You write a rule once and define where it applies. Reflect delivers it to your agent at that step of the run, shows which conversations it reached, and measures whether it helped. It also surfaces the mistakes your agent keeps making, so you know which rule to write next. ### Why not put the rule in my system prompt or a skill? A rule in the system prompt rides along on every request, so rules pile up and each one gets buried among the rest. A skill is read only when the model decides to open it, so the rule can be missed. Reflect delivers each rule where you said it applies, keeps every rule in one place, scoped to the right accounts and workflows, and shows whether it helped. ### Does Reflect guarantee my agent follows the rule? No. Reflect makes sure the rule reaches your agent where you said it applies, and shows you the conversations it reached. Whether the model then follows it is what the measurement is for: you compare outcomes with and without the rule on real runs. ### How is Reflect different from evals and observability tools? Observability shows what your agent did. Evals check it against test cases before you ship. Reflect controls what your agent is told in production: it delivers your rules where they apply, shows which conversations each one reached, and measures each one on real runs. It works alongside both. ### How do I connect my agent? Install the Reflect SDK, send your agent’s conversations, and request the rules that apply once per user turn. It works with any agent framework and any model, and your app keeps control of its model calls and tools. ### Does Reflect add latency? Reflect answers in 200 ms on average, once per user turn. It fails open: if a response takes longer than 750 ms, your agent continues without the rules. ### Can I check a rule before it goes live? Yes. Reflect previews each rule against your stored conversations, so you see every one it would have reached, including any you didn’t expect, before you turn it on. ### How much does Reflect cost? Every plan includes unlimited agents. Hobby is free with 1,000 turns a month shared across your agents, and you can sign up without talking to anyone. Growth is $299 a month plus $0.005 per turn. Enterprise is custom pricing for high volume. ## Company and contact - Email: support@usereflect.dev - [Y Combinator profile](https://www.ycombinator.com/companies/reflect) - [LinkedIn](https://www.linkedin.com/company/runreflect) - [X](https://x.com/reflectdotdev) - [YouTube](https://www.youtube.com/@reflectdotdev) ## Optional - [Privacy](https://usereflect.dev/privacy) - [Terms](https://usereflect.dev/terms)