AI Workspaces: PewDiePie's Odysseus and Clanker Cloud

An AI workspace brings your conversations, files, instructions, and tools together so an assistant can work on a project across multiple tasks. The practical question is what you can get back: a report you can send, research you can check, a cleaned dataset, or a small working App.

Clanker Cloud Workspace is a browser-based workspace for that work. Projects hold related files, instructions, and outputs. Sessions are where you ask for work and review progress. Agents bring the instructions, knowledge, Skills, and selected Tools for the job. You can begin with a normal browser account and a few source files.

What does an AI workspace do?

The value is keeping enough context to do the next task well. A client report needs the current notes and the agreed reporting format. A vendor comparison needs proposals, selection criteria, and evidence. A spreadsheet workflow needs column definitions and rules for missing values.

Job	Inputs	Result to inspect
Client update	Meeting notes, current brief, previous report	Progress, decisions, blockers, and next steps
Vendor research	Proposals and selection criteria	Evidence table, unanswered questions, recommendation
Data cleanup	CSV and column definitions	Cleaned copy, change log, reconciled totals
Small work App	Sample spreadsheet and workflow description	Private preview tested against expected results
Reusable specialist	Instructions, reference knowledge, Skills, Tools	An Agent tested on a representative task

Start with an output you know how to judge. Add more tools or recurring execution after the first result works.

What is PewDiePie's Odysseus?

Odysseus is a self-hosted AI workspace associated with PewDiePie. Its current repository describes chat, agents, research, documents, email, notes, calendar, and local model workflows. The original pewdiepie-archdaemon/odysseus repository URL now redirects there.

The documented Docker setup runs the workspace on your own host. This approach suits people who want to operate a self-hosted environment and local model workflows. Read the project's current setup and security documentation before installing it; deployment and model configuration are part of the work.

Where Clanker Cloud fits

Clanker Cloud Workspace lets you start project work in a browser. You can research, create reports and other files, configure custom Agents, build Apps, and set up recurring work. The Workspace overview and FAQ explain the current capabilities and their boundaries.

For a closer look at the interface, take the Clanker Workspace tour or try a first task from notes to a saved file. Both walk through how the conversation, Project context, and reviewed output fit together.

For example, a small business owner can keep a supplier brief and proposals in one Project, ask an Agent for a comparison, review the evidence, and save the final decision. A consultant can keep meeting notes, reporting instructions, and last week's output together for the next client update.

If you already use Claude or another compatible assistant, connect it to Workspace so it can work with approved Project tools and files. For a repeatable business job, build a hosted Agent and choose an email or recurring trigger after testing the task once.

Clanker Cloud also supports DevOps, connected computers, and the open-source Clanker CLI. Those are additional paths when the job needs infrastructure or computer access. Ordinary browser tasks do not require local cloud credentials or a CLI installation.

Self-hosted or browser-based: what should you evaluate?

Decision	Odysseus	Clanker Cloud Workspace
Getting started	Install and operate the documented self-hosted setup	Open Workspace in a browser and create or sign in to an account
Useful first test	Try the task on your chosen host and model configuration	Try a report or research task with Project files and inspect the output
Operations	Plan for host setup, updates, storage, and model configuration	Review account access, selected Tools, hosted data handling, and usage
Costs to inspect	Hosting, hardware, and your configured model costs	AI work and cloud-environment credits; current allowances and optional plans

Use the same source material and review criteria when evaluating either approach. Read each product's current documentation for the exact features you need. This is a comparison of starting workflows, not a benchmark of result quality.

A worked example: notes into a client report

Here is a fictional first task you can judge without connecting outside services. Create a Project for Northstar Studio, start a Session, and add these notes as a file or paste them with your request:

Northstar Studio — weekly notes
Reporting date: 13 October 2026
Website draft is ready for client review.
Decision: launch only after the client approves the draft.
Blocker: product images have not arrived.
Maya will send the image request on 14 October.
Client approval date has not been confirmed.

Ask your Agent:

Create a one-page client report from the Northstar Studio notes.
Use sections for progress, decisions, blockers, and next steps.
Keep names and dates exactly as provided. Mark missing dates as
unconfirmed. Do not invent a launch date or claim approval was given.
Save the report as a Project file and show where I can open it.

An illustrative result would include:

Progress: Website draft is ready for client review.
Decision: Launch requires client approval.
Blocker: Product images are missing.
Next step: Maya sends the image request on 14 October.
Unconfirmed: Client approval date and launch date.

This example is fictional; it is not a measured customer run. Check your actual result against the notes: is the owner right, is the date right, and did the Agent keep approval and launch unconfirmed? Request corrections before downloading or sharing the report.

The weekly client report template includes a fuller prompt with fictional notes, downloadable starter text, and a review checklist. Opening Workspace does not automatically import the starter or send its prompt; copy it into a Session yourself.

Reuse what worked

Keep reviewed source files and results in the Project. Add fresh notes for the next report and identify which inputs are current. To reuse the reporting method, build a custom Agent with instructions, relevant knowledge, Skills, and selected Tools.

Saved context helps with later tasks. It does not automatically retrain the model or schedule future runs. For recurring work, test a task once, then configure a Loop or Schedule and review its timing, access, history, and ongoing costs.

Common questions

Do I need the desktop app?

You can start Workspace in your browser. Work on your own computer needs the desktop app and an approved connection to that device. Cloud browsers, desktops, and hosted compute run separately.

Can I use this for everyday business work?

Yes. Begin with a client report, a vendor comparison, or a spreadsheet workflow. Specify the inputs, expected output, and what must be checked.

What does it cost to try?

New accounts start with $20 in included credits. AI work and active cloud environments use credits at their applicable rates. Recurring workers have ongoing costs. Review current pricing and credit usage, then check the result and usage of your first task.

What happens to my files?

Projects keep related files, instructions, and outputs together. Choose which sources and Tools an Agent can use. Hosted work processes data through the applicable cloud services and model providers; review the Privacy Policy and Subprocessor Register for retention and data handling.

Sources

• Odysseus repository and current README: self-hosted scope and setup, reviewed 13 October 2026.
• Clanker Cloud Workspace: browser workflows and product overview.
• Workspace FAQ: Projects, Sessions, Agents, access, and recurring work.
• Weekly client report template: fictional inputs and review criteria.
• Pricing: current included credits and usage terms.

Open Workspace and try one task whose result you can check.