Guide: Give Hermes Cross-Device Memory with Hindsight Cloud

If you run Hermes on more than one machine, cross-device memory is the difference between one assistant and three disconnected ones. Without a shared memory backend, your laptop Hermes, desktop Hermes, and server Hermes each build their own partial history. You end up re-explaining the same context everywhere.

Hindsight Cloud fixes that. Point multiple Hermes installs at the same memory backend, keep a stable bank strategy, and the assistant carries context across devices instead of starting over.

Quick answer

1. Set up Hermes with Hindsight Cloud.
2. Use the same bank IDs on every device that should share memory.
3. Keep project or user bank naming stable.
4. Verify recall from one device after retaining facts on another.
5. Use separate banks when you want isolation.

Why local memory is not enough across devices

Local memory works until your workflow spans machines.

Maybe you research on a laptop, code on a desktop, and run automation on a server. If each environment keeps memory locally, then your context fragments immediately. The information exists, but it is scattered.

Cross-device memory solves that by moving the bank behind the agent instead of inside one machine.

Step 1: Choose Hindsight Cloud

The easiest path is the Hermes memory setup wizard:

hermes memory setup

Choose Hindsight, then choose Cloud.

That gives every Hermes install a shared backend instead of a machine-local one.

Step 2: Make your bank strategy explicit

A shared backend is only half the job. The other half is making sure every device points at the same bank when it should.

Good patterns:

• one bank per user assistant
• one bank per project
• one bank per customer or account

Examples:

• ben-personal
• project-hindsight-docs
• acct-acme

The key is stability. If your laptop uses project-hindsight-docs and your desktop uses docs-project, you did not create cross-device memory. You created two separate histories with similar names.

Step 3: Verify the flow end to end

A quick verification loop:

1. On device A, tell Hermes one fact worth remembering.
2. End the session.
3. On device B, start a fresh Hermes session using the same bank.
4. Ask for the fact.

If the second device can recall it cleanly, the setup is working.

Best use cases for cross-device memory

Personal assistant workflows

You might message Hermes from your phone, continue from your laptop, and later ask a follow-up from your desktop. Shared memory makes those all feel like one ongoing relationship with the same assistant.

Coding workflows

Research on one machine, implementation on another, deployment or monitoring on a server. A shared bank keeps the repo context continuous.

Team operations

A server-side Hermes worker can build memory that a human-operated Hermes chat can later use, and vice versa.

When not to share a bank

Not everything should be in one pool.

Use separate banks when:

• different projects should stay isolated
• personal context should not mix with team memory
• customer-specific memory should not leak across accounts

Cross-device should not mean cross-everything.

Common mistakes

• using different bank IDs on different machines
• reusing one giant bank for unrelated work
• assuming Cloud alone guarantees good retrieval without a clear naming strategy
• forgetting to test recall from a second device

FAQ

Can I share one bank across laptop and server?

Yes. That is one of the main reasons to use Cloud.

Do I need the same local Hermes install path on every machine?

No. What matters is the shared backend and stable bank IDs.

Should teams share one bank too?

Only when the workflow itself should share memory. Otherwise, keep banks separate.

Next Steps

• Read move Hermes memory from local files to Hindsight Cloud
• Review single bank vs multi-bank with Hindsight
• Start with Hindsight Cloud