Your project’s files, shared with your team and every AI they use.
Put the brief, the documents and the tasks in one project. Open it in Claude, Claude Code, Codex, or ChatGPT on plans that allow it. Whatever your AI writes or edits is saved back to the project under your name, where everyone sees it.
One project and up to five people writing to it, free. No card.
Or set it up with us on a callFrom an imagined team, written for this page.
A new site and a launch for Kiln Coffee’s spring blend, in cafés from 15 October.
What it changes comes back here, under your name, for everyone on the project.
- Send the interview summary to Kiln CoffeeTomorrowresearchThe project agent proposes it due Friday: Evren is away until ThursdayAcceptDismiss
- Check the FAQ against the briefwas due yesterdaylanding
- Book the launch newsletter slotSundaylanding
- Draft the launch announcementlandingyouthe project agent
- Edited Landing copy v3
- Left a note on FAQ
- Added Competitor scan to research
- Proposed a change to Send the interview summary
- Joined the project
How it works.
- 1
Bring your project in.
The brief, Word and PDF files, a whole folder, or a GitHub repository.
- Brief
- Kiln brand guide.pdf
- Interview notes.docx
- kiln-sitelinked from GitHub
- 2
Open it from your AI.
Add Alcoven to Claude, or ChatGPT on plans that allow it. Or clone the project for Claude Code and Codex. Each person uses their own AI, or the app.
In Claude or ChatGPThttps://app.alcoven.co/mcpHow to add it in ClaudeIn Claude Code or Codexgit clone https://app.alcoven.co/git/your-project.git - 3
Your AI’s edits are saved to the project.
Edit a document or add a task from your AI, and your name is on it. Your teammate’s AI reads it the next time it opens the project.
EKEmre Kaya, with Claude Code2 min agoEdited Landing copy v3 from git
DADefne Aksoy1 h agoAdded the task Send the interview summary for Emre from Claude
Working from your own AI costs nothing extra on any plan, because we run no model for it.
How it goes today.
Your AI remembers your own chats. When your team uses different AIs, what one of them writes reaches the others only when someone copies it across.
The paste
The brief goes into a new chat. Again.
The relay
Summarised, forwarded, pasted into another AI.
The wrong client
Another client’s colours turn up in this project.
How it compares.
People ask us most about Claude and ChatGPT projects, and about GitHub. They are good tools, and you can keep using them: Claude opens Alcoven projects, as does ChatGPT on plans that allow it, and a GitHub repository can be linked into one. Here is what happens in each, in five moments every team knows.
| A Claude or ChatGPT project | A GitHub repository | Alcoven | |
|---|---|---|---|
| A teammate uses a different AI | They join with an account on your AI, or work from a summary someone sends them | Their coding agent works in it through git; in chat, their AI can only read it | They open the same project from their own AI, and what their AI changes is saved there too |
| Your AI finishes a piece of work | A person saves or uploads it to the project; in ChatGPT, teammates can also read the chat | Claude Code or Codex commits it and you push; from a chat, you copy it in yourself | Your AI saves it to the project under your name, and every teammate’s AI reads it there, whichever AI they use |
| You’re back after two days off | You ask someone, or go back through the chats you can see; in ChatGPT, you can also ask it about the project’s chats | The commit history shows every change and who made it; you work out which ones are new to you | You see what changed since you last opened the project, and who changed it, or ask your AI |
| The client wants to follow along | They need an account on that AI; in Claude, a seat in your organisation | They need a GitHub account | Invite them by email; readers join free |
| The team moves to another AI | You set the project up again in the new one | Nothing moves; git works with any coding agent | Nothing moves; open the same project from the new AI |
ChatGPT opens Alcoven projects on plans where an MCP tool can write.
Inside a project.
Documents and files in folders, tasks for people and agents, and an agent that answers from the project. Every change shows who made it.
Every file, and what your team changed since you last opened it.
The brief and every document, in folders, with every past version. Open one and you see what others changed since you last read it, and who changed it.
- Landing copy v32 h agoHero, launch date, tasting boxEmre
- FAQYesterdayWhere is it roasted?Emre
- Hero photo notes3 days agoTwo shots from the roasteryyou
- Launch newsletterLast weekSubject lines to tryyou
Meet Hearth, our spring blend.
In cafés from 15 October.
1 October → 15 October
Order a tasting box before launch day.
FAQ
Where is it roasted? In small batches, at the Kiln roastery.
Agents propose. You decide.
People and agents share the task list, and an agent can take a task with a person as its owner. When an agent thinks a task belongs to someone else or another day, it proposes the change and says why. Nothing moves until a person accepts.
An agent proposed these. They wait for you.
- Send the interview summary to Kiln CoffeeTomorrowAccepted · UndoresearchThe project agent proposes it due Friday: Evren is away until ThursdayAcceptDismiss
- Check the FAQ against the briefwas due yesterdaylanding
- Book the launch newsletter slotSundaylanding
- Draft the launch announcementlandingyouthe project agent
Ask the project. Approve its edits.
The agent in the app answers from the project, and can search the web. When a document should change, it proposes the exact edit and why. Nothing changes until you approve it, and you can undo it.
One place doesn’t. The brief renamed the blend Ember on Friday, and the hero in §Landing copy v3 still says Hearth.
What it saves you.
No re-briefing.
The new chat reads the brief from the project.
Fewer catch-up meetings.
The client asks where things stand. You answer from the tasks and documents in the project.
New people catch up on their own.
A new freelancer reads the whole project on day one.
No redos from the wrong client.
What the other client wanted stays in the other client’s project.
Free for one project.
One project and up to five people writing to it, free. No card.
- Readers and clients join freeYou pay for each person who can write to a project. Readers, clients and agents join free, on any plan.
- Your own AI costs nothing extraWorking from your own AI, over git or MCP, costs nothing extra on any plan, because it runs no model on our side.
- Agent use, in dollarsThe agent in the app comes with the plan: $1 of use a month for the team on Free, $3 per person writing on Solo and Team.
Questions.
Plans in full are on the pricing page. Anything else, write to info@alcoven.co.
What is Alcoven?
A shared place for a project’s brief, documents, files and tasks. Each person on the team opens it in the app or from their own AI (Claude, Claude Code, Codex, or ChatGPT on plans where an MCP tool can write), and whatever their AI changes is saved to the project under their name. Every past version is kept, and every project is also a git repository you can clone.
How is it different from a Claude or ChatGPT project?
A shared Claude or ChatGPT project gives everyone on it the same files and instructions to start a chat from, inside that one AI. In Claude each person’s chats stay their own unless they share one; in ChatGPT members can see each other’s chats, and ChatGPT can draw on them. Either way, what a chat produces reaches the project’s files only when someone saves or uploads it, and a teammate who works in another AI can’t open the project from there (how sharing works in Claude and in ChatGPT). In Alcoven your AI edits the project’s documents and tasks directly, the change is saved under your name, and teammates see it from Claude, ChatGPT, Codex or the app. Anyone you invite by email can join, whichever AI they use, and clients read free.
How is it different from a GitHub repo?
An Alcoven project is a git repository: Claude Code and Codex clone it and push to it. It is also open to the people on the team who never use git. They open it in Claude or ChatGPT, or in the app, where Word files, PDFs and spreadsheets open in a preview and tasks sit beside the documents. Clients join free as readers. Your code can stay on GitHub: link a repository and it shows up as a folder in the project, and changes to it go back to GitHub as pull requests.
What does it cost?
Free covers one project and up to five people writing to it, with no card. Solo is $8 a month, or $80 a year, for one person writing on any number of projects. Team is $12 per person writing a month, billed yearly, or $15 per person writing, billed monthly. Prices are in US dollars. See pricing.
Who do I pay for?
You pay for each person who can write to a project. Readers, clients and agents join free, on any plan.
What happens when the agent’s included use runs out?
The agent in the app pauses until next month, unless the team adds more. Any team, Free included, can add $10, $20 or $50 of agent use in advance: $10 of agent use costs $11, plus any tax for your country, and what the team adds carries over from month to month for 12 months. On Solo and Team, the team can also have it added automatically, up to a monthly limit it sets. Nothing is billed afterwards. Working from your own AI never counts, because we run no model for it.
Can I use it from Claude, ChatGPT or Codex?
Yes. Clone the project with git and work on it from Claude Code, Codex or any coding agent, then push. Or add Alcoven to an AI app that speaks MCP, such as Claude, or ChatGPT on plans where an MCP tool can write, and choose the project when you sign in (how to set it up in Claude). What your AI changes comes back under your name, and it costs nothing extra on any plan.
Does everyone on the team need the same AI?
Each person opens the project from the AI they already use, or in the app, and everyone works from the same files. If you move from ChatGPT to Claude, the project stays as it is: open it from Claude instead.
Is this another project management tool?
It holds what a project knows: the brief, the documents, the tasks and every change, with who made it. Tasks are open or done, and there are no boards to groom.
Who can see a project?
The people and agents you add, each allowed to read, write or manage. Other projects stay out. Nothing is public.
Is my project used to train a model?
We don’t train models. The agent in the app runs on Amazon Bedrock, and AWS says that neither AWS nor the model providers use what goes into or comes out of Bedrock to train any model (Amazon Bedrock FAQs). When you work from your own AI, that tool’s own terms apply.
What if I want to leave?
Clone the project with git, on any plan: every document, task and past version comes with it. Cancel any time. Your projects stay readable, and you can still clone them. You can ask for a refund within 14 days of a payment.
I work alone. Is it for me?
Yes. One project is free, and Solo covers any number of projects for one person writing. When you add a person later, they start with everything the project holds.
Who is behind Alcoven?
Deep Node Studios Yazılım ve Teknoloji Limited Şirketi, in Kadıköy, İstanbul. Paddle handles payments as our merchant of record. Write to us at info@alcoven.co.
The last time you explain this project.
Put it in once. Everyone on the team, and every AI they use, works from the same files.
One project and up to five people writing to it, free. No card.
Or set it up with us on a call