🔍 Read the full analysis: How A Shared Desk Changed My AI Agent Setup 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 AI-run publishing network is moving editorial work from chat and a code-focused task tracker to Thrymvault, a shared workspace for story status, drafts and editor feedback. The system is designed to record which approved application made changes and to prevent agents from deleting content. The article describes the workflow, but some features and backup plans are still being rolled out.
Thorsten Meyer says his AI-run publishing network is shifting editorial work from chat and a code-focused task tracker to Thrymvault, a shared workspace where agents can update story status, submit drafts and receive editor feedback. The change is intended to make work across StrongMocha News Group easier to follow and audit, although Meyer’s November 13 report distinguishes the system’s design from features that are already running.
Meyer describes a day this month when agents helped steer nine articles for his sites, gathering sources, drafting copy and preparing graphics. The workflow relied on a chat window and a task tracker built for code work. To check progress, Meyer had to ask in chat; headline changes depended on the right agent seeing and acting on a message; and a completed explainer page could not be shared as a link, he wrote.
Thrymvault is a workspace for pages, databases and boards that Meyer’s team says it 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 guide for agents is generated from the same tool catalogue as the server, according to the report, so its instructions correspond to the available tools.
In the proposed editorial flow, a topic appears as a board card. Meyer decides whether it moves forward; moving it from “Topic” to “Kit” signals that an editorial agent should gather primary sources and check figures. The agent then drafts the story, updates its status and posts a report alongside the card. Meyer can request a revision in the card’s thread by mentioning the relevant application. The workflow ends with a scheduled and published link on the card, or a “stopped” status that tells agents to leave the story alone.
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
A Traceable Desk for Agent Work
The development addresses a practical coordination problem as AI agents handle multiple parts of publishing in parallel: instructions, status and revisions can become difficult to track when they are scattered across chats. A shared board could give an editor one place to see what is underway and attach requests to the story they concern.
Meyer also presents traceability as an editorial safeguard. In the system he describes, changes carry the name of the approved application that made them, displayed as “via” that application. Agents cannot delete content, and access is limited to applications approved by a workspace owner. Those controls could help an editor identify how a change entered a draft and respond when a correction is needed. They do not, by themselves, establish that content is accurate or that every change will be caught.
The approach is relevant to publishers and other teams using agents for shared work: it treats access, review and records as part of the workflow, rather than relying on conversational instructions alone. Meyer’s account is a description of his own system and newsroom, not independent evidence that the setup improves accuracy or output across organizations.
Products Worth Considering
LG UltraGear 32GX850A-B 32-inch 4K UHD OLED Gaming Monitor
Samsung Odyssey OLED G5 27-Inch QHD Gaming Monitor
Logitech MX Mechanical Wireless Illuminated Keyboard, Tactile Quiet, Graphite
As an affiliate, we earn on qualifying purchases.
From Chat Threads to Story Cards
Meyer’s report identifies the immediate trigger as work on nine articles, during which he had to ask agents for progress and send revisions through chat. He argues that chat handles a single conversation more readily than an ongoing project with a story, sources, graphics and a deadline. In his account, messages can scroll out of view, agents may interpret them differently, and a chat alone does not provide a clear record of who changed what.
Thrymvault is presented as a desk rather than a replacement for the editorial decision-maker. The editor chooses which topics proceed and can stop work; agents handle research and drafting under application-level access. Meyer says decisions that change agent behavior, such as a headline rule, will be recorded in a separate Decisions database with the wording, date, scope and a link to the rule it changes. The rule itself remains where agents can read it.
The report says the workspace runs on the group’s own hardware. A second machine kept on standby and hourly backups are described as a next step, not a live feature. That distinction matters because the article covers both an intended newsroom process and infrastructure that has not yet been deployed.
“When I wanted to know where a story stood, I had to ask.”
— Thorsten Meyer, describing the newsroom’s previous workflow
What Is Live and What Is Planned
The report does not provide an independent audit of Thrymvault’s security, reliability or editorial outcomes. It describes controls including approved applications, keys limited to one workspace, application attribution and a ban on agent deletions, but does not establish how those controls have been tested or how exceptions are handled.
It is also not clear which parts of the described editorial flow are already in regular use and which remain under rollout. Meyer explicitly says a standby machine and hourly backups are not live yet. The report does not give a deployment schedule for those measures, adoption figures, comparative results against the prior workflow or evidence that the new setup reduces factual errors.
Meyer says the next infrastructure step is adding a second machine on standby and hourly backups. The report does not give a launch date. In the newsroom workflow he outlines, the next work is to use the board to move stories from topic selection through source gathering, drafting, editor feedback and publication, with decisions and changes recorded along the way.
Further reporting would be needed to establish which functions are now live, how often agents use the workspace, and whether the records and access limits work as intended in day-to-day publishing. The available account is Meyer’s description of the design and current rollout, rather than an external evaluation.
Key Questions
What is Thrymvault?
Thrymvault is a self-hosted workspace for pages, databases and boards that Meyer says his team built for people and AI agents to coordinate work.
How would agents use the newsroom board?
Agents can update a story card with its status, draft link and open questions, and attach reports to the card. An editor can request changes in the card’s thread by mentioning the relevant application.
What controls does Meyer describe for agent access?
According to Meyer, agents enter through owner-approved applications, changes are labeled with the application that made them, and agents cannot delete content. He also describes keys restricted to one workspace and no more rights than the application permits.
Are backups and a standby machine already in place?
No. Meyer describes a second machine on standby and hourly backups as the next step, not as features already live.
Has Thrymvault been shown to improve newsroom accuracy?
The report does not provide comparative accuracy or reliability results. Meyer explains the intended workflow and audit trail, but their effect on factual errors has not been established in the source material.
Source: ThorstenMeyerAI.com
