🔍 Read the full analysis: Can AI Agents Work From A Shared Desk? on ThorstenMeyerAI.com
Get smart everyday buys delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
TL;DR
Thorsten Meyer says his StrongMocha News Group newsroom is building Thrymvault as a shared workspace for human editors and AI agents, replacing chat-only coordination with boards, pages and traceable changes. The system’s agent access controls and integrations are described, but a standby machine and hourly backups are still planned rather than live.
Thorsten Meyer has outlined how his StrongMocha News Group newsroom is using Thrymvault, a self-hosted workspace designed to let AI agents and a human editor manage story work in one shared place. The system is intended to replace chat-only coordination with editorial boards, attached reports and changes attributed to approved applications; Meyer says a standby machine and hourly backups are planned but not yet running.
Meyer said that, earlier this month, his team guided nine articles through a chat window and a task tracker designed for code work. Agents gathered sources, drafted copy, prepared graphics and assembled drafts. But the editor had to ask for status updates, submit headline changes through chat and rely on the correct agent to notice them. Meyer also said he could not share one completed explainer as a link.
Thrymvault is a workspace for pages, databases and boards that Meyer’s group built and hosts on its own hardware. Its agent access currently includes an MCP server with 20 tools, a REST API and a command-line tool. According to the article, agents can enter only through applications approved by a workspace owner; their changes are identified by the application used, and agents cannot delete content.
In the proposed newsroom workflow, an editor places a topic on a board and decides whether it advances. Moving a card to a “Kit” stage signals an editorial agent to collect sources and check figures; the agent then drafts and updates the card with its status, draft link and open questions. Editors can request revisions in the card’s thread by mentioning the relevant agent application. The card can later carry the published link or be marked stopped. Meyer says decisions that change agent instructions will also be recorded in a database linked to the relevant rule.
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 Workflow for AI Stories
The proposal addresses a practical editorial problem: when several agents work in parallel, a chat transcript can leave the status of a story, its current instructions and responsibility for changes scattered across messages. A shared board could give an editor a single view of work in progress and keep corrections tied to the story they concern.
For readers, Meyer’s stated rationale is that a record of who drafted material, who requested a change and which rule applied could make errors easier to investigate and correct. That is a design goal, not evidence that the system has already improved accuracy or reduced mistakes. The article provides no comparative results on story quality, speed or error rates.
The access model also matters for organizations using agents on editorial or other collaborative work. Meyer describes owner-approved applications, restricted permissions and attribution of changes as safeguards. Those controls may make activity easier to review, while leaving the human editor responsible for deciding what proceeds and what is published.
Popular right now
Ninja Foodi 7-Qt MegaZone FlexBasket Air Fryer, 6-in-1, Black, DZ071
Ninja Foodi 10-Qt DualZone XL Air Fryer, 6-in-1, Gray (DZ401)
Bose QuietComfort True Wireless Earbuds with Active Noise Cancellation, Bluetooth 5.3, USB-C Charging, IPX4, Black
As an affiliate, we earn on qualifying purchases.
From Chat Threads to Editorial Cards
Meyer frames Thrymvault as an operational response to a specific newsroom experience, rather than as a general-purpose claim about AI publishing. He says chat worked for individual exchanges but proved less useful when a story involved sources, graphics, deadlines and several agents. In his account, instructions could scroll out of view, agents could interpret a message differently, and it was difficult to see who had changed what.
The workspace has a familiar pages-and-databases format, Meyer writes, but its focus here is the agent interface and the group’s self-hosted setup. The agent guide is published both as plain text and as a machine-readable manifest; Meyer says both are generated from the same tool catalogue as the server. The intended result is for an agent’s instructions to match the tools actually available to it.
The article distinguishes current functions from planned infrastructure. Thrymvault runs on the group’s own machines, while a second machine on standby and hourly backups are described as next steps, not active services. The source material does not give a rollout schedule for those additions.
“When I wanted to know where a story stood, I had to ask.”
— Thorsten Meyer, describing the newsroom’s chat-based workflow
Rollout and Results Still Undocumented
The article describes the system’s design and some capabilities, but does not provide a detailed list separating every feature already in use from every feature still being rolled out. It also does not state when the planned standby machine and hourly backups will be operational.
No independent evaluation or performance figures are provided. It remains unclear how often agents miss comments, how permissions are tested, how the newsroom handles incorrect or conflicting edits, and whether the workflow has changed the time needed to publish or correct a story. The source describes safeguards and intended processes; it does not establish that those measures prevent errors.
Meyer’s account concerns his own organization and setup. The article does not report adoption by other newsrooms or provide a comparison with commercial collaboration platforms.
Backup Infrastructure and Ongoing Rollout
Meyer identifies a second machine on standby and hourly backups as the next infrastructure steps, while making clear that neither is live yet. The article gives no launch date. In the newsroom workflow, agents are expected to continue updating story cards and responding to editor comments through their approved applications, with cards carrying a published link or a stopped status when work ends.
Further reporting would be needed to establish how widely the workflow is in use, which described functions are fully deployed, and whether the record of agent activity improves editorial oversight in practice. For now, the confirmed development is Meyer’s account of a system being built and used within his network, not a reported industry-wide adoption.
Key Questions
What is Thrymvault?
Thrymvault is a workspace for pages, databases and boards that Thorsten Meyer says his group built and hosts on its own machines. This account focuses on its use by newsroom staff and AI agents.
How do AI agents access the workspace?
Meyer says agents can use an MCP server with 20 tools, a REST API or a command-line tool. Access is limited to applications approved by a workspace owner, according to the article.
Can agents delete newsroom content?
Meyer says agents cannot delete anything in the described setup. The article does not provide an independent security review of that control.
Are backups and a standby machine already in place?
No. Meyer describes a second machine on standby and hourly backups as planned next steps, not live features.
Does the article show that the system improves accuracy?
No comparative accuracy or error-rate results are reported. Meyer argues that a traceable record could make corrections easier, but the article presents that as a rationale for the design rather than a measured outcome.
Source: ThorstenMeyerAI.com
