🔍 Read the full analysis: My Approach To A Shared Workspace For AI Agents on ThorstenMeyerAI.com
Get business pricing on tech for your team
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
Thorsten Meyer says his StrongMocha News Group is building Thrymvault as a shared workspace where newsroom agents can update story boards, drafts and reports. The design uses owner-approved applications and records which application made each change; backup hardware is planned, not yet running.
Thorsten Meyer has outlined Thrymvault, a self-hosted workspace built for AI agents and editors at StrongMocha News Group, saying it is intended to replace scattered chat instructions and code-focused task tracking with shared story boards and traceable changes. Meyer’s account describes the system’s design and some functions already in use, while noting that a standby machine and hourly backups are planned but not live.
Meyer said that on one day this month his network guided nine articles through a chat window and a task tracker designed for code work. Agents gathered sources, wrote drafts, created graphics and prepared copy. But the editor had to ask for each story’s status, send headline changes through chat and rely on the intended agent seeing them. Completed explainer pages also could not be shared as links, he wrote.
Thrymvault is a workspace for pages, databases and boards that Meyer’s group built and hosts on its own machines. Agents can access it through an MCP server with 20 tools, a REST API or a command-line tool. The agent guide is generated from the same tool catalogue as the server, Meyer said, so instructions and available tools are meant to stay aligned.
In the newsroom workflow Meyer describes, a topic appears as a card for editorial review. Moving it to a “Kit” stage signals an editorial agent to assemble sources and check figures; the agent then drafts the story and records status, links and open questions on the board. Editors can request changes in comments that mention an agent’s application. The workflow ends with the published link attached to the card, while a stopped card tells agents to leave the work alone. Meyer says moving a card is an instruction agents follow, not an automatic software switch.
The shared desk · animated
How a story moves through the shared desk
One card per story. The editor moves it and asks questions in its thread; the agent does the work, keeps its status current and answers there. Every agent change is signed with its application.
Editorial board
Thread on the card
Editor@Editorial agent shorter title, no colon
AgentDone: title shortened, draft updated.
via Editorial agent
Signed history
- Moved to Kit · by the editor
- Status and draft link updated · via Editorial agent
- Comment mentions Editorial agent · by the editor
- Reply in the same thread · via Editorial agent
- Approved and scheduled · by the editor
- Published link added · via Editorial agent
Agents add and edit. They cannot delete anything.
Source: Thrymvault agent integration guide and rollout plan, November 2026. Shows the desk as designed for StrongMocha News Group; the board templates and the pick-up of mentions are still being rolled out.
Animation: Thorsten Meyer AI
The proposal addresses a coordination problem created when several agents work on the same reporting pipeline: a chat message may not show the current state of a story, who is responsible for the next step or whether a correction was made. A shared board could give editors a single place to review progress and attach instructions to the relevant assignment.
For readers, the potential value is traceability. Meyer says changes are labeled with the approved application that made them, using a “via” marker, and that editorial decisions can be recorded with dates, scope and links to the rules they change. Such a record could help a newsroom investigate and correct mistakes. The source material describes the design goal; it does not provide independent evidence that the system has improved accuracy, speed or reader outcomes.
The project also reflects a broader operational choice: agents are not treated as anonymous users with unrestricted access. Meyer says each enters through an application approved by a workspace owner, and agents cannot delete content. That model may matter to organizations seeking to use automated tools while retaining human editorial control.
Products Worth Considering
Apple MacBook Air 15-inch Laptop with M5 Chip, 16GB Memory, 512GB SSD, Midnight
Logitech MX Master 4 Ergonomic Wireless Mouse with Haptic Feedback, Graphite
Bose TV Speaker (2nd Gen) – Compact Soundbar with Bluetooth and HDMI‑ARC Connectivity, Clear Dialogue, Dolby Atmos, Deep Bass, Remote Contro
As an affiliate, we earn on qualifying purchases.
From Chat Threads to Story Cards
Meyer’s article frames the workspace around a specific experience at his AI-run network, rather than a general product launch. Chat remained useful for direct instructions, but he says it did not provide a persistent, shared status view across stories, sources, graphics and deadlines. A task tracker built for code work also did not give him the newsroom-oriented way to review work he wanted.
Thrymvault is compared in broad shape to Notion, but its stated distinction here is that it runs on the publisher’s own hardware and offers agents limited access through dedicated interfaces. The article presents a design for an editorial desk, not a claim that every described workflow is fully deployed. It separates functions Meyer says are available from work still being rolled out, including the planned backup arrangement.
““That is the point where an AI-run newsroom outgrows chat.””
— Thorsten Meyer, writing on ThorstenMeyerAI.com
Deployment and Safeguards Still Developing
The source does not give a detailed rollout schedule or a full list of which newsroom features are already operating in production. It says a second machine on standby and hourly backups are the next step and not live yet. It also does not report independent security testing, reliability data or results comparing the workspace with the previous chat-based process.
Some safeguards are described at a high level. Meyer says access is limited to applications approved by a workspace owner, that a key is restricted to one workspace and has no more rights than its application allows, and that agents cannot delete anything. The supplied article text cuts off during its description of key permissions, so the precise permission model and other access controls are not fully available here. It is also unclear how consistently agents follow board instructions, including a request to stop work.
Backups and Rollout Remain Ahead
Meyer identifies a standby machine and hourly backups as upcoming work, but gives no launch date. He says the article distinguishes what runs today from features still being rolled out; the material provided does not include a complete feature-by-feature deployment list.
The next evidence readers can look for is a fuller account of which workflows are in routine use, how the permission and audit controls operate in practice, and whether the backup plan has gone live. Until then, Thrymvault is best understood as a described newsroom workspace and operating approach, with some active components and some development still underway.
Key Questions
What is Thrymvault?
Thrymvault is a self-hosted workspace for pages, databases and boards that Meyer says is designed for editors and AI agents at StrongMocha News Group.
How would agents use the newsroom workspace?
Meyer says agents can work through an MCP server, REST API or command-line tool. In his proposed editorial flow, they update story cards, attach draft links and report open questions.
Can an AI agent publish or delete a story on its own?
The source says agents cannot delete content and describes an editor deciding which topics move forward. It does not fully explain publishing permissions or establish that every described workflow is active.
Are hourly backups already running?
No. Meyer describes a second machine on standby with hourly backups as a next step, and says that setup is not live yet.
Has Thrymvault been shown to improve newsroom accuracy?
The article explains the intended benefits of shared status and traceable changes, but provides no measured results showing effects on accuracy, speed or reader outcomes.
Source: ThorstenMeyerAI.com
