A fit for developer-tool companies whose docs lag each release, and for open-source maintainers. It replaces the chore of scanning merged pull requests for pages to edit, while a person still approves the result. A customer story on its site says Vellum has it drafting over 50% of doc updates.
What changed
Promptless is an AI agent that turns merged code, Slack threads and tickets into cited documentation pull requests, with support for 11 named Git-backed docs platforms including Mintlify and Docusaurus. Writers review each draft in a dashboard before anything publishes, and every suggestion links back to the source that prompted it.
Maker: Promptless · Autonomy: semi autonomous · Maturity: GA
Underlying models: Model-agnostic (AWS Bedrock by default)
About Promptless
Promptless is an AI agent that keeps customer-facing documentation in sync with what a software team ships. It was founded in San Francisco by Prithvi Ramakrishnan, who was previously VP of Product and Engineering at the fintech infrastructure company Bond, and Frances Liu, who led product and machine learning engineering at Cloudflare, OneSignal and Graft. It is a Y Combinator Winter 2025 company. Unlike a writing assistant that waits for a prompt, it wakes up on its own when a pull request opens, a Slack thread about docs starts or a documentation ticket is created, decides whether a doc change is warranted and drafts it.
The workflow has four stages: listen, draft, review and publish. Triggers include changes on GitHub and GitLab, chat messages, support tickets and API calls. While drafting, it researches the context you connect, such as code, Confluence, Notion, Linear or Jira, then attaches citations to each suggestion so a reviewer can check the claim against its source. By default every suggestion waits in a dashboard for a person to refine it. Teams can switch on automatic pull requests, and automatic merging is a separate opt-in policy that skips protected branches. It reads your existing pages to match their style and can run the Vale prose linter before a suggestion reaches your queue.
It suits developer-tool companies and open-source projects whose documentation falls behind each release. Any Git-backed platform works, including Mintlify, GitBook, Docusaurus and MkDocs. The vendor links public pull requests in the Vitess and Helm docs repositories, where project maintainers review each change. Related tools do adjacent jobs: kapa.ai answers questions from docs that already exist, Swimm keeps code walkthroughs in sync, and Dosu captures engineering knowledge for coding agents. Review agents such as Greptile, CodeRabbit and GitHub Copilot act on the same pull requests, but they check code rather than write documentation.
A second product, Promptless for Agent Instructions, applies the same loop to the Skills, subagents, hooks and AGENTS.md files that steer coding agents. It analyses session traces from agents such as Claude Code, Gemini CLI, OpenClaw, Cursor and Devin, then proposes fixes you review and merge. A hosted MCP server also lets those same editors submit documentation tasks to the docs product.
Cloud customers run on AWS Bedrock by default, and the vendor says it never uses customer data to train or fine-tune models. It keeps no copy of source code or support conversations, only your documentation and the feedback you give. A self-hosted deployment with your own model setup is available. The product ships monthly: September 2026 added direct page edits from Claude over MCP and a beta LaunchDarkly trigger, and October 2026 added Cursor as a trace source.
Pricing
Read on promptless.ai/pricing on 24 October 2026. The docs product is priced by site size in normalized pages, with flat monthly plans and unlimited contributors. Startup covers small sites on Slack and GitHub; Growth bundles scale from $500 up to $4,000 a month by page count and add Jira, Linear, Confluence, Notion, translations and screenshot updates. Enterprise is custom. Agent-instructions plans are quoted on a demo call. Every plan includes a free trial.
Key Features
- Event-driven triggers: A merged GitHub or GitLab change, a Slack or Microsoft Teams thread, a Jira or ClickUp ticket, or an API call can start a docs update with no prompt.
- Cited suggestions: Each draft carries inline citations to code functions, Slack conversations, websites or support tickets, so reviewers verify claims against the source.
- House-style matching: Promptless learns your existing pages, accepts a style guide, and runs Vale on the prose it writes, fixing findings at or above your configured alert level.
- Screenshot capture: Promptless Capture regenerates product screenshots, with crops and annotations, when the UI changes; it is part of the Growth plan.
- Internal context lookup: It searches Confluence spaces, Notion databases, Linear issues, Slite notes and Google Drive (beta) so docs follow the written product spec.
- Hosted MCP server: Claude Code, Codex, Cursor and VS Code connect over OAuth, and a coding agent can open an account that a person claims within 72 hours.
- Agent-instruction governance: It analyses agent session traces and proposes reviewed changes to Skills, subagents, hooks and AGENTS.md files, with an Impact page showing recovered time.
- Publishing policies: Suggestions wait in the dashboard by default, with optional automatic pull requests and a separate auto-merge setting.
Strengths
- Open Promptless work in the Vitess and Helm documentation repos, reviewed by project maintainers, lets you judge real output before you buy.
- It runs on whichever Git-backed docs platform you already use, so adopting it means no docs migration.
- Citations on every suggestion turn review into a source check rather than a rewrite, which is the main reason reviewers can trust drafts quickly.
- The vendor states it does not train models on customer content and keeps no copy of source code, which helps in a security review.
Weaknesses
- The cheapest published docs plan is a flat fee tied to site size, so a very small docs site pays for capacity it may not use.
- Growth prices are ranges that depend on page count, and Enterprise and the agent-instructions product have no public price.
- The pages we read offer customer quotes but no accuracy or acceptance-rate benchmark for the docs product.
- Cloud processing runs on AWS by default, so teams that need their own model setup must arrange a self-hosted deployment with the vendor.
Frequently Asked Questions
What does Promptless actually cost?
The docs product is priced by the size of your documentation, counted in normalized pages, so a long API reference counts as several. Startup is a flat $500 a month for up to 200 pages. Growth bundles climb through four page bands to $4,000 a month for up to 5,000 pages, and Enterprise is custom with unlimited pages. Agent-instructions plans have no public price.
Is Promptless free to use?
Not for companies: there is no free plan, but every paid plan begins with a 14-day trial. The exception is the open-source program, which the vendor says is free for CNCF, Linux Foundation and other eligible non-commercial projects. Two small utilities, Agent Score and Broken Link Report, are also free.
What should you use instead of Promptless?
Swimm fits teams that want auto-synced documentation of their own codebase for engineers. Dosu suits teams that want engineering knowledge from GitHub, Slack and Notion shared with coding agents over MCP. kapa.ai answers customer questions from docs you already maintain, so it needs a writer or a tool like Promptless upstream.
What separates Promptless from Dosu?
They overlap on syncing docs but start from different ends. Dosu is built around a knowledge layer that coding agents query, and its listing shows a free plan limited to public repositories. Promptless is built around customer-facing pages, publishes a page-count price, and keeps a writer in the review seat. Pick Promptless when a human ships the docs.
How do you set up Promptless?
Sign up at accounts.gopromptless.ai, install the GitHub App on your docs repository, connect Slack, and set triggers in a promptless.yaml file. Suggestions then appear in the dashboard for review. If you work from an editor, add the MCP server with a single command in Claude Code, Codex, Cursor or VS Code.
Top Alternatives
- Swimm: Pick Swimm if your docs are internal code walkthroughs for engineers, Promptless if the pages are customer-facing.
- Dosu: Pick Dosu if the goal is feeding coding agents what engineers learned, Promptless if the output is docs a writer ships.
- kapa.ai: Pick kapa.ai to answer support questions from docs you already have, Promptless to write and update those docs.