# Simular Platform (Sai API) > Sai is Simular's computer-use agent, available to developers as the Sai API. Give it a task in plain English and it does the work on a real computer the way a person would, by seeing the screen, clicking and typing: on a Windows or Linux cloud computer that Simular hosts, or on your own Mac or Windows PC. Your agent can then get things done where there is no API: web apps behind a login, admin consoles, desktop software, forms with no import. Run a workflow once and Sai can save it as code: later runs replay that code and ask the model only where the screen matters, which cuts tokens by up to 90% on repetitive tasks, and the agent takes over when the UI changes. On OSWorld 2.1, Sai has the highest partial score on the official leaderboard, 79.38%, and the only OSWorld 2.1 entry that publishes its cost, $14.34 per task, with about 1.5x fewer model calls than models used on their own, Oct 2026. Coding agents (Claude Code, Codex, Cursor) use it through the `@simular-ai/sai-mcp` MCP server. The free plan includes a cloud computer. ## Key facts - Product: the Sai API, the computer-use agent API from [Simular](https://www.simular.ai/). Portal, docs and pricing: https://platform.simular.ai. - Interfaces: `POST /v1/agents/message` over HTTPS, a Python SDK, and the `@simular-ai/sai-mcp` MCP server for Claude Code, Codex, Cursor and any stdio MCP client. - Computers: a Windows or Linux cloud computer that Simular hosts, or your own Mac or Windows PC. - Models: model-agnostic. You choose the model Sai plans with. - Cost: a workflow Sai has done once can be saved as replayable code, which cuts tokens by up to 90% on repetitive tasks. - Reliability: saved code heals itself. When a step breaks because the UI changed, the agent takes over from there. - Benchmark: on the official OSWorld 2.1 leaderboard, Sai has the highest partial score, 79.38%, at $14.34 per task. On binary accuracy it is second, at 41.67% (Oct 2026). - Safety: Sai asks before sign-ins, credentials, phone checks and other sensitive steps. Link-only approvals go to the person as an approval link. - Pricing: Explore is free, with daily credits and limited access to a Windows cloud computer. Pay as you go is $50 a month, Enterprise custom. See [https://platform.simular.ai/pricing](https://platform.simular.ai/pricing). - Research: built on Agent S, Simular's open-source agentic framework (ICLR 2025), and the Neuro-Symbolic Computer Use paper (arXiv, Sep 2026). ## When to use Sai instead of writing code Use Sai when: - There is no API, or the API doesn't cover the task: vendor portals, admin consoles, legacy web apps. - The task needs a real browser session with someone's accounts. Sign-ins, passwords and phone checks go to the human as an approval link. - The work happens in a desktop app, on Windows, macOS or Linux. - It is a one-off, and a script would cost more to write and keep working than the task itself. - You need to see what a real user sees: testing a sign-up flow, reproducing a bug report. - It is a workflow you will repeat: run it once, then replay it as code (below). Write code yourself when: - A documented API or SDK does the job. - The data is already local or in the repo. - You need exact, byte-for-byte output, such as file formats or data transforms. ## Run once, replay as code Most computer-use agents re-plan every step on every run. Sai can turn a workflow it has done into a reusable skill: the steps that stay the same become code, and the model is asked only about what depends on the screen. 1. Run the task once with Sai. 2. In the same conversation, ask: `Save this as a replayable skill named invoice_export, with vendor and month as parameters.` 3. Next time, ask: `Run invoice_export for vendor Acme, month March.` Sai replays the code, and when a step breaks because the UI changed, the agent takes over from there. ## Example tasks - `Open our staging site in Chrome, sign up with the test Gmail account, click the verification email, and tell me where the flow breaks.` - `Log in to the vendor portal (it has no API) and list last month's invoices with their totals.` - `Enter each row of the uploaded CSV into the supplier form at https://example.com/suppliers/new, which has no import, and report the rows that failed validation.` - `Follow the steps in this bug report in Chrome on Windows and describe exactly what happens at each step.` - `In the admin console, export the user list and tell me which accounts haven't signed in for 90 days.` ## Getting started 1. Sign in with Google at [https://platform.simular.ai/sign-in](https://platform.simular.ai/sign-in) and create an API key at [https://platform.simular.ai/api-keys](https://platform.simular.ai/api-keys). Keys start with `sapi_` and are shown once. Free accounts verify a phone number before their first key. 2. Add Sai to your coding agent as an MCP server, replacing `sapi_YOUR_KEY` with the key. Node.js 22.12 or newer is required. 3. Ask the agent for a task, for example: `Use Sai to open Notepad on my cloud computer, type "Hello from Sai", and tell me the window title.` ### Claude Code ```bash claude mcp add sai -e SAI_API_KEY=sapi_YOUR_KEY -- npx -y @simular-ai/sai-mcp ``` Optional: install the Sai skill, which teaches Claude Code the task loop. ```bash npx -y @simular-ai/sai-mcp init-claude ``` ### Codex ```bash codex mcp add sai --env SAI_API_KEY=sapi_YOUR_KEY -- npx -y @simular-ai/sai-mcp ``` ### Cursor Add this to `~/.cursor/mcp.json` (or `.cursor/mcp.json` in a project), then restart Cursor. The same block works for any stdio MCP client. ```json { "mcpServers": { "sai": { "command": "npx", "args": [ "-y", "@simular-ai/sai-mcp" ], "env": { "SAI_API_KEY": "sapi_YOUR_KEY" } } } } ``` ## Notes for coding agents - You cannot create the API key yourself. Ask the user to create one at https://platform.simular.ai/api-keys and give it to you, then put it in `SAI_API_KEY` as shown above. - After adding the server, the user may need to restart the client or reload its MCP servers before the tools appear. - Tools: `sai_machines`, `sai_models`, `sai_task_start`, `sai_task_wait`, `sai_task_approve`, `sai_task_abort`, `sai_upload`. - Direct control tools: `sai_control_open`, `sai_control_exec`, `sai_control_interrupt`, `sai_control_close`. These let the coding agent operate a cloud computer and inspect screenshots. Read [direct control](https://platform.simular.ai/documentation/coding-agents/direct-control.md) and the [control HTTP API](https://platform.simular.ai/documentation/api-reference/control.md) before use. Missing MCP tools after setup do not prevent using HTTP directly. - Live viewing: return the task's `watch_url`. Codex's inline view cannot stream; use that link instead. Read [live viewing](https://platform.simular.ai/documentation/coding-agents/live-view.md) and the [live-screen HTTP API](https://platform.simular.ai/documentation/api-reference/machines.md). The MCP viewer and `watch_url` are view-only. For a person to use the mouse and keyboard, call `sai_live_link` and return its `live_url`. For sign-ins and other link-only human input, return the actual approval link. - Direct control addresses a `machine_id`, not a task's `session_id`. Closing its lease deletes a free-plan computer; it is not a handoff that preserves the screen for the user. - The loop: `sai_task_start` returns a `session_id`. Call `sai_task_wait` with it, passing back the `cursor`, while `status` is `running`. `idle` means `text` holds Sai's answer. - On `needs_approval`, decide with `sai_task_approve`. When the approval is link-only (sign-ins, credentials, phone checks), give `approval_url` to the user and keep waiting. - Tasks run on a real computer and may act on the user's accounts. Confirm anything irreversible with the user first. ## Debug with curl The raw request behind every task. It streams events until `finish`. If Sai asks before a step, the stream sends `data-approval-request` and waits: when `isLinkOnly` is `false`, answer it from another terminal with `POST /v1/agents/approve` (`{ "approvalId": "…", "response": "yes" }`, or `"task"` to allow the rest of the task). Only link-only approvals are completed at `approvalUrl`. ```bash curl -N https://api.simular.ai/v1/agents/message \ -H "Authorization: Bearer $SAI_API_KEY" \ -H "Content-Type: application/json" \ -d '{"message": "Open Notepad, type Hello from the Sai API, and tell me the window title"}' ``` ## FAQ ### What is the Sai API? The Sai API lets a program or a coding agent hand a task to Sai, Simular's computer-use agent, in plain English. Sai does the task on a real computer by seeing the screen, clicking and typing, and returns the result. ### How is Sai different from a browser automation API? A browser automation API drives a web browser. Sai operates the whole computer, so one task can move between websites, desktop apps and file dialogs, on Windows, Linux or macOS. ### Which computers does Sai run on? A Windows or Linux cloud computer that Simular hosts, or your own Mac or Windows PC. On Pay as you go, up to 5 cloud computers can run at the same time. ### How does Sai cut the cost of repeated tasks? Run a workflow once, then ask Sai to save it as a replayable skill. Later runs execute the saved code and ask the model only where the screen matters, which cuts tokens by up to 90% on repetitive tasks. ### How does Sai handle sign-ins and other sensitive steps? Sai pauses for approval before sensitive steps. Sign-ins, credentials and phone checks are link-only approvals: Sai sends the person an approval link, and they complete the step themselves. ### How does Sai score on OSWorld? On the official OSWorld 2.1 leaderboard, Sai has the highest partial score, 79.38%, at $14.34 per task, the only published cost on the leaderboard. On binary accuracy it is second, at 41.67%. Details: [Sai on OSWorld 2.1](https://platform.simular.ai/benchmark/osworld-benchmark). ### How much does the Sai API cost? Explore is free, with daily credits and limited access to a Windows cloud computer. Pay as you go is $50 a month, and Enterprise is custom. See [pricing](https://platform.simular.ai/pricing). ### Does Sai work with Claude Code, Codex and Cursor? Yes, through the `@simular-ai/sai-mcp` MCP server. One command adds it; see Getting started above. ## Docs - [Introduction](https://platform.simular.ai/documentation): what the Sai API is and how a request flows through it - [Quick start](https://platform.simular.ai/documentation/quickstart): from an API key to a finished task - [Claude Code](https://platform.simular.ai/documentation/coding-agents/claude-code): add Sai to Claude Code, plus the optional Sai skill - [Codex](https://platform.simular.ai/documentation/coding-agents/codex): add Sai to OpenAI Codex - [Cursor](https://platform.simular.ai/documentation/coding-agents/cursor): add Sai to Cursor or any stdio MCP client - [Approvals](https://platform.simular.ai/documentation/concepts/approvals): how Sai asks before sign-ins, credentials and other sensitive steps - [Billing and limits](https://platform.simular.ai/documentation/concepts/billing-and-limits): what a request costs, the free plan and rate limits - [API reference](https://platform.simular.ai/documentation/api-reference/message): `POST /v1/agents/message`, the request behind every task ## Research - [Neuro-Symbolic Computer Use](https://arxiv.org/abs/2609.36927): the research behind replay. Learned policies cut per-run cost 15–217× and latency 3.4–5.1×, and raised Pass^3 by 3.6–15.8 points over the base agent on OSWorld-Verified and ScienceBoard (Sep 2026) - [Sai on OSWorld 2.1](https://www.simular.ai/articles/sai-tops-osworld-2-0): the highest partial score on the official leaderboard, 79.38%, and the only OSWorld 2.1 entry that publishes its cost, $14.34 per task, with about 1.5x fewer model calls than models used on their own, Oct 2026 - [Agent S](https://www.simular.ai/articles/agent-s): the open agentic framework behind Sai (ICLR 2025) - [Agent S on GitHub](https://github.com/simular-ai/Agent-S): source and papers; Agent S3 reached 72.6% on OSWorld (TMLR 2026) ## Benchmarks - [Sai on OSWorld 2.1: Leaderboard and Cost per Task](https://platform.simular.ai/benchmark/osworld-benchmark): Sai has the highest partial score on the official OSWorld 2.1 leaderboard, 79.38%, at $14.34 per task. See the full leaderboard, trajectories and method. ## Use cases - [Use Cases for Sai, the Computer-Use Agent](https://platform.simular.ai/use-cases): Tasks Sai does on a real computer: behind logins, inside desktop apps and across tools with no API. Pick a role, start from a prompt, run it from the API. - [Test Your README Install Steps on Fresh Windows](https://platform.simular.ai/use-cases/test-readme-install-on-windows): Sai follows your README on a clean Windows cloud computer in a visible PowerShell window, like a new user, and reports the steps and warnings the docs leave out. - [Compare AI Tool Pricing From Official Pages](https://platform.simular.ai/use-cases/compare-ai-tool-pricing): Compare ChatGPT, Claude, Gemini, Cursor, Copilot, Midjourney and more from their own pricing pages: cheapest paid plan, features and limits in one table. - [Automate a Legacy Windows App With No API](https://platform.simular.ai/use-cases/automate-legacy-windows-app): Windows desktop automation API for agents: Sai works legacy apps from the screen, with no UI Automation hooks, asks before risky steps and replays as code. - [Automate Data Entry: CSV Rows Into a Web App, No Import](https://platform.simular.ai/use-cases/automate-data-entry-from-csv): Automate data entry into web apps with no import or API. Sai fills the form for each CSV row, skips duplicates, reports failed rows and replays as code. - [Test Your Windows Installer on a Clean Cloud Computer](https://platform.simular.ai/use-cases/test-windows-installer): Installer testing on a clean Windows machine: Sai clicks through UAC, SmartScreen and the setup wizard, then checks first launch, upgrade and uninstall. - [User Access Review Automation for Apps Without an API](https://platform.simular.ai/use-cases/user-access-review-automation): User access review automation for SaaS apps with no API or SCIM: Sai exports users from admin consoles, flags 90-day inactive accounts and saves evidence. - [Desktop App Testing from Claude Code, No Scripts](https://platform.simular.ai/use-cases/test-desktop-app-from-claude-code): Desktop app testing from Claude Code: Sai installs your latest build on a cloud computer, clicks through key flows and reports failures with screenshots. - [How to Reproduce a Bug Report in a Real Browser](https://platform.simular.ai/use-cases/reproduce-bug-report): Hand Sai a ticket or issue. It follows the steps in a real browser with a test account and returns exact steps, screenshots and whether the bug reproduced. - [Log Calls in Your CRM From Notes and Transcripts](https://platform.simular.ai/use-cases/update-crm-from-notes): Log calls in your CRM from notes or a transcript: activity, next step, close date, custom fields and follow-up tasks. Sai asks before changing deal stage. - [Meeting Notes to Action Items in Your Tracker](https://platform.simular.ai/use-cases/meeting-notes-to-action-items): Turn meeting notes into action items with an owner and due date, create them in Asana, Jira, Linear or Trello, and post a Slack recap after you approve. - [Research Brief in Your Template, Every Claim Cited](https://platform.simular.ai/use-cases/web-research-brief): Turn a question into a one-page research brief in Google Docs or Word, with a citation for every claim, using the web and the sources behind your logins. - [Track Flight Prices and Hotel Rates for One Trip](https://platform.simular.ai/use-cases/track-flight-prices): Track flight prices and hotel rates daily across Google Flights, airline, Kayak and hotel sites. Sai logs each price and alerts you below your target. - [MCP Computer Use for Claude Code, Codex, Cursor](https://platform.simular.ai/use-cases/gui-steps-from-coding-agents): Your coding agent hands the steps it cannot reach, such as admin consoles and desktop apps, to Sai over MCP. Sai does them and reports back. - [Computer Use API as One Tool for Your AI Agent](https://platform.simular.ai/use-cases/computer-tool-for-your-agent): Add a computer use API to your agent as one tool in LangGraph, the OpenAI Agents SDK or the AI SDK, for apps with no API. No VM or screenshot loop needed. - [Runbook Automation for Web UI Tasks, No Scripts](https://platform.simular.ai/use-cases/runbook-to-scheduled-job): Runbook automation for steps that live in web dashboards: describe the runbook once, watch the first run, then schedule it. Sai asks before risky steps. - [Compare Prices Across Websites, Shipping Included](https://platform.simular.ai/use-cases/compare-prices-across-sites): Compare prices across websites by delivered total: shipping, tax, member prices and coupons on each store. Sai lists the cheapest option and never buys. - [Restaurant Reservation Bot That Follows the Rules](https://platform.simular.ai/use-cases/restaurant-reservation-bot): Not a reservation bot that grabs and resells tables. Sai uses Resy Notify and OpenTable alerts, checks your calendar and asks before anything is booked. - [Auto Fill Online Forms, Uploads and PDFs](https://platform.simular.ai/use-cases/auto-fill-online-forms): Auto fill online forms from your saved details and documents: multi-page applications, uploads and PDFs. Sai stops before submit for your review. - [Cancel Unwanted Subscriptions You Forgot (2026)](https://platform.simular.ai/use-cases/cancel-unused-subscriptions): Find recurring charges across cards, PayPal and app stores, then cancel unwanted subscriptions on each company's own page. You approve every cancellation. - [End-to-End Testing for Signup Flows, No Scripts](https://platform.simular.ai/use-cases/end-to-end-test-signup): Test your signup flow end to end like a new user: form, verification email, onboarding and paid plan in a real browser. Results and issues, no scripts. - [Release Notes from Pull Requests, Written for Users](https://platform.simular.ai/use-cases/release-notes-from-prs): Turn merged pull requests into release notes your users can read. Sai reads each PR, groups what changed and publishes to GitHub, Notion and Slack. - [Convert a Document to PowerPoint, Fully Editable](https://platform.simular.ai/use-cases/document-to-slides): Turn a Word doc, PDF or Google Doc into a slide deck built in PowerPoint or Google Slides with your template, native charts and notes. Fully editable. - [Monitor Competitor Website Changes Every Week](https://platform.simular.ai/use-cases/monitor-competitor-websites): Monitor competitor website changes on pricing, product and changelog pages. Sai sees what a buyer sees, explains each change and posts a weekly digest. - [Download Bank Statements Automatically (2026 Guide)](https://platform.simular.ai/use-cases/download-bank-statements): Download bank statements automatically from every bank portal, every month. Compare bank feeds, statement APIs and AI agents, then set it up in Sai. ## Comparisons - [Compare the Sai API With Other Tools](https://platform.simular.ai/compare): Side-by-side comparisons of the Sai API with Claude, OpenAI and Gemini computer use, Browser Use, Browserbase and Cua, plus roundups of RPA and desktop automation, AI agent builders and APIs, MCP servers, GUI automation, testing clouds and AI sandbox tools. - [Sai API vs Claude Computer Use (2026)](https://platform.simular.ai/compare/claude-vs-simular-api): Sai API vs Claude Computer Use: a ready computer-use agent, or a model toolset you host and wire up yourself. Setup, OSWorld 2.1, cost and safety. - [Sai API vs OpenAI Computer Use (2026)](https://platform.simular.ai/compare/openai-vs-simular-api): Sai API vs OpenAI’s computer tool: a ready computer-use agent, or a model you run in your own loop and sandbox. Setup, benchmarks, cost and safety. - [Sai API vs Codex Computer Use (2026)](https://platform.simular.ai/compare/codex-vs-simular-api): Sai API vs Codex Computer Use: an API agent on a cloud or your own computer, or a Codex desktop-app feature on your own machine. Setup, cost and fit. - [Sai API vs Gemini Computer Use (2026)](https://platform.simular.ai/compare/gemini-vs-simular-api): Sai API vs Gemini Computer Use: a ready computer-use agent, or a Gemini model tool you run in your own browser, VM or device. Setup, cost and safety. - [Sai API vs Browser Use (2026)](https://platform.simular.ai/compare/browser-use-vs-simular-api): Sai API vs Browser Use: a computer-use agent for websites and desktop apps, or a hosted web-browsing agent on cloud Chromium. Setup, cost and fit. - [Sai API vs Browserbase (2026)](https://platform.simular.ai/compare/browserbase-vs-simular-api): Sai API vs Browserbase: a computer-use agent for websites and desktop apps, or cloud headless browsers with the Stagehand SDK. Setup, cost and fit. - [Sai API vs Hyperbrowser (2026)](https://platform.simular.ai/compare/hyperbrowser-vs-simular-api): Sai API vs Hyperbrowser: a computer-use agent for websites and desktop apps, or cloud browsers with hosted browser agents and scraping APIs. Cost and fit. - [Sai API vs Cua (2026)](https://platform.simular.ai/compare/cua-vs-simular-api): Sai API vs Cua: a hosted computer-use agent you call with one task, or open-source sandboxes and drivers you build your own agent on. Setup and cost. - [Best RPA Software in 2026: 10 Tools Compared](https://platform.simular.ai/compare/best-rpa-software): Compare the best RPA software in 2026: UiPath, Power Automate, Automation Anywhere, Blue Prism, open-source RPA and AI agents, with prices checked. - [Playwright MCP Alternatives: 8 Browser MCPs Compared](https://platform.simular.ai/compare/playwright-mcp-alternatives): Compare Playwright MCP alternatives in 2026: Chrome DevTools MCP, Playwright CLI, Browser MCP, Browserbase, Browser Use, Selenium MCP and Sai, with setup. - [GUI Automation Tools: PyAutoGUI & pywinauto Alternatives](https://platform.simular.ai/compare/gui-automation-tools): Compare GUI automation tools in 2026: pywinauto, PyAutoGUI, SikuliX, AutoHotkey, AutoIt, FlaUI, nut.js and AI agents, with versions and licenses checked. - [Best AI Sandboxes for Agents in 2026: 8 Compared](https://platform.simular.ai/compare/ai-sandboxes): Compare AI sandboxes for agents: E2B, Daytona, Modal, Vercel Sandbox, Cloudflare, Cua, Browserbase and full computers, with prices and limits checked. - [UiPath Alternatives in 2026: 8 Competitors Compared](https://platform.simular.ai/compare/uipath-alternatives): Compare UiPath alternatives and competitors in 2026: Power Automate, Automation Anywhere, Blue Prism, Robot Framework, Appian, SAP, IBM and AI agents. - [Best Computer Use MCP Servers in 2026: 8 Compared](https://platform.simular.ai/compare/computer-use-mcp-servers): Compare computer use MCP servers for Claude Code and Cursor: Windows-MCP, Claude Code computer use, Cua Driver, Peekaboo, Terminator and Sai, with installs. - [Best Open-Source Computer Use Agents in 2026 | Agent S4, UI-TARS](https://platform.simular.ai/compare/open-source-computer-use-agents): Compare open-source computer use agents: Agent S4, UI-TARS Desktop, OpenCUA, Holo4, Microsoft UFO, Cua and Browser Use, with licenses and OSWorld results. - [Best MCP Servers for Claude Code in 2026: 10 Compared](https://platform.simular.ai/compare/best-mcp-servers-for-claude-code): The best MCP servers for Claude Code: GitHub, Context7, Playwright, Chrome DevTools, Sentry, Figma, Linear, Supabase, Notion and Sai, with install commands. - [WinAppDriver Alternatives in 2026: 9 Tools Compared](https://platform.simular.ai/compare/winappdriver-alternatives): WinAppDriver has had no stable release since v1.2.1. Compare NovaWindows, Appium WinCore, FlaUI, pywinauto, TestComplete, Ranorex, Katalon, Tosca and Sai. - [BrowserStack Alternatives in 2026: 9 Compared](https://platform.simular.ai/compare/browserstack-alternatives): Compare BrowserStack alternatives on price and coverage: Sauce Labs, TestMu AI (LambdaTest), TestingBot, AWS Device Farm, Playwright Workspaces, Perfecto and more. - [Best Desktop Automation Tools in 2026: 9 Compared](https://platform.simular.ai/compare/desktop-automation-tools): Compare desktop automation software for Windows and Mac: Power Automate, UiPath, AutoHotkey, Keyboard Maestro, Automation Anywhere, pywinauto, AI computer use and the Sai API. - [Best AI Agent Builders in 2026: 10 Compared](https://platform.simular.ai/compare/best-ai-agent-builders): Compare AI agent builder platforms, no-code to code: n8n, Gumloop, ChatGPT workspace agents, Copilot Studio, Zapier Agents, Lindy, Dify, Google, LangGraph and Sai API. - [Best AI Agent APIs in 2026: 9 Compared](https://platform.simular.ai/compare/best-ai-agent-apis): Compare AI agent APIs and agent-as-a-service platforms: OpenAI Agents API, Claude Managed Agents, Gemini, AgentCore, Foundry, Browser Use, Manus, Devin and Sai API. ## Cookbooks - [Cookbooks: build with the Sai API](https://platform.simular.ai/cookbooks): End-to-end builds on the Sai API: from an API key to a working agent that uses a real computer, with the prompts, commands and code. - [Track Competitor Ads in the Meta Ad Library with Claude Code and the Sai API](https://platform.simular.ai/cookbooks/track-competitor-ads-meta-ad-library): Have Claude Code send Sai to the Meta Ad Library on its own cloud computer, collect the active ads for three sneaker brands, and turn them into a clean CSV and a scored ad breakdown. - [Build a Personal AI Agent in 10 Minutes with the Sai API](https://platform.simular.ai/cookbooks/build-your-own-personal-agent): Get a Sai API key, give Codex or Claude Code a cloud computer over MCP, and ship a personal AI agent with chat, approvals and a live screen. ## Optional - [All Sai API docs for LLMs](https://platform.simular.ai/documentation/llms.txt): every docs page in one index, with raw Markdown copies - [Simular](https://www.simular.ai/): the company behind Sai