Hermes Lets You Move a Local Agent to the Cloud in One Step from Desktop
A Hermes agent built on your own computer can now be moved to the cloud in full. A new Backup & Transfer section in Desktop settings either makes a backup or replaces the cloud agent with the local one. It carries over memory, skills, chats, profiles, and scheduled tasks, and an Undo button reverts a mistake. That makes it practical to debug the agent at home and roll out a finished copy to the cloud.
AI-processed from X: @HermesWatcher; edited by Hamidun News
Anyone who has seriously set up an AI agent knows that its most valuable part isn't the code but the accumulated context. Memory of your preferences, a set of skills, chat history, profiles for different tasks, and scheduled jobs build up over weeks. Losing them when you move from a laptop to a server is painful, and rebuilding them takes a long time. The Hermes team has addressed this problem: the Desktop version now includes a feature for moving a local agent to Hermes Cloud.
What Was Added to Desktop
The news came from @HermesWatcher, an account that tracks the project through a daily summary of changes. That day, November 16, 54 pull requests were merged into the project, but the change most useful to practitioners was the Backup & Transfer section in the Desktop app's settings.
It has two modes. The first is a regular backup: you save the agent's entire configuration and can return to it later. The second is replacement: the local agent becomes the new content of an agent that already exists in Hermes Cloud. Instead of rebuilding the setup by hand, you choose the action in settings, and the cloud version takes on the same "identity" as the one at home.
The transfer includes five groups of data: memory, skills, chats, profiles, and scheduled tasks. The developers also added an Undo option that lets you roll back the operation. For replacement, this is an important safeguard: you are overwriting a working cloud agent, so a mistake costs more than it would with a simple backup.
Why It Matters for Developers and Product Teams
The main practical effect is a more convenient agent lifecycle. Previously, a typical workflow looked like this: you experimented locally, where it was safe to break settings, and then had to reproduce the result in the production environment. Now the local computer can serve as a sandbox and the cloud as production, with a tested copy moving over.
This is especially useful in three situations:
- you are refining an agent's skills and want to see how it behaves at home before updating the cloud agent;
- the agent has already built up useful memory that cannot be lost when the environment changes;
- you have scheduled tasks, such as regular reports or checks, that should keep running in the cloud without reconfiguration.
For teams building agents for clients, the time savings are also significant: a reference configuration can be assembled once and then transferred, and the backup serves as a restore point.
What to Watch Out For
The original announcement does not detail the specifics, so it is worth checking a few things before the first transfer. First, back up the cloud agent before replacing it: even with Undo, an extra backup costs nothing. Second, confirm that the integrations and access the agent uses in the cloud still work after the move. The summary lists memory, skills, chats, profiles, and tasks, but says nothing about keys and external services. Third, note that replacement overwrites the cloud agent rather than merging the data of the two versions. If valuable memory has also accumulated in the cloud, it may be lost.
There is also a broader takeaway for the market. Agent platforms increasingly resemble familiar development tools: they are gaining environments, backups, rollbacks, and transfers between them. While an agent lives on a single machine, it cannot be treated as a reliable part of a product. The ability to move an agent's state cleanly between local and cloud environments brings these systems closer to standard engineering practice.
Other changes in the same summary, such as a plugin for local models on Mac, document reading, and surfacing reasoning output in Telegram, show that the project is moving fast. But moving an agent to the cloud addresses the most down-to-earth problem: not losing what has been accumulated when leaving the experimental stage for real work.
- Nov 21, 2026VibeCBS collects project nicknames in advance: the agent finds 2.8 times more documents
- Nov 21, 2026FDE job postings grew 4.6x in a year: what a forward deployed engineer actually does at the client's site
- Nov 21, 2026Sakana AI Opens Applied Research Lab for Market and Economic Modeling
- Nov 21, 2026Meta Is Reportedly Preparing the $300 Muse Charm AI Tamagotchi With a Year of Verizon 5G
Key Questions Answered
What Was Added to Desktop?
The news came from @HermesWatcher, an account that tracks the project through a daily summary of changes. That day, November 16, 54 pull requests were merged into the project, but the change most useful to practitioners was the Backup & Transfer section in the Desktop app's settings. It has two modes. The first is a regular backup: you save the agent's entire configuration and can return to it later. The second is replacement: the local agent becomes the new content of an agent that already exists in Hermes Cloud.
Why It Matters for Developers and Product Teams?
The main practical effect is a more convenient agent lifecycle. Previously, a typical workflow looked like this: you experimented locally, where it was safe to break settings, and then had to reproduce the result in the production environment. Now the local computer can serve as a sandbox and the cloud as production, with a tested copy moving over.
What to Watch Out For?
The original announcement does not detail the specifics, so it is worth checking a few things before the first transfer. First, back up the cloud agent before replacing it: even with Undo, an extra backup costs nothing. Second, confirm that the integrations and access the agent uses in the cloud still work after the move. The summary lists memory, skills, chats, profiles, and tasks, but says nothing about keys and external services.
Build & Ship Production AI Agents Without Boilerplate Code
Start your free 72-hour trial at Hamidun Academy (academy.hamidun.com) — interactive lessons on Claude Code & Cursor, live Tech Radar, and production agent templates.
The AI world, distilled — once a week
Seven stories that actually mattered, hand-picked. No noise, no reposts, no press releases.
Done! Check your inbox for a confirmation.