HappytoHelp

Support is here.

What might software look like in 2 years? We tried to imagine it, then built it. AI customer support, MIT, 100% free.

Run
Sculpted toy rider with raised spear beside a lantern
Two travelers crossing a stream on a narrow wooden plank, rendered as a matte sculpture.
02

Help before they need it.

Put the right knowledge in reach before a small question becomes a roadblock. Bring your help content, customer context and conversations together, so the next useful step is easier to find.

Give customers a clear way forward with answers grounded in your own knowledge. When they need more help, your team can step into the conversation with the context already there.

The process

  1. Connect knowledge

    Bring help content, customer context and conversations together.

  2. Ground answers

    Use published knowledge, with source citations people can follow.

  3. Bring in your team

    Keep the conversation and its context together when a person steps in.

Built from your sourcesYour help contentCustomer contextConversationsPublished knowledge
03

Keep watch as your site changes.

Your product keeps moving. New pages go live, features change and yesterday’s answer can fall out of date. Connect your website and help content so your support knowledge can move with them.

Refresh content on a schedule that suits your team, and review new versions before they replace published knowledge. Keep useful answers connected to what your customers actually see.

  1. Published

    Your current, usable knowledge.

  2. Refresh

    Bring in new or changed source content.

  3. Review & publish

    Check the new version before replacing published knowledge.

A cloaked traveler keeping watch from a stone window, rendered as a matte sculpture.

A failed refresh keeps the last usable published version in place.

04

Your agents, working together.

Bring the agents you already work with into your support workflow. Connect Codex or Claude through your local computer, and give them the relevant context to help with work that reaches beyond a single reply.

Choose what they can access and which tasks they can take on. Your team stays in control of the handoff, with work kept in context and replies ready to review.

  1. Pair a computer

    Connect Codex or Claude on a local computer.

    Your computerScoped permissions
  2. CodexClaude

    Let agents investigate

    Give the task relevant context and a defined scope.

    Codex · ClaudeAccess you choose

  3. Review the reply

    Keep your team in control of what happens next.

    Human handoffRevoke access
One traveler reaching down to help another up a rocky climb, rendered as a matte sculpture.
05

100% free. MIT licensed.

Run it. Change it. Make it yours. HappyToHelp is built around a simple idea: the software your team relies on should be software you can own and adapt.

Keep the code, shape the experience and run it in accounts you control. Choose the hosting and AI providers that fit your needs, with the freedom to keep improving the product as your team grows.

  • MIT licensed

    Use, change and keep the source code.

  • Your hosting

    Run it in accounts you control.

  • Your AI provider

    Choose the services that fit your team.

View the source ↗
Two travelers sharing water at a stone well, rendered as a matte sculpture.
What is free?

The source code. Run and change it under the MIT license. Hosting and AI usage are paid through the accounts you choose.

Can my team still reply?

Yes. Keep the conversation and its context together when a person steps in.

Where do agents run?

Connect a local computer running Codex or Claude, choose the work it can do, and revoke access when you need to.