Use case / multiple agents

Hermes Agent Bot Mode: Multiple Agents, One Runtime

One assistant doing every job becomes vague fast. Hermes Agent bot mode lets you split work across named bots, so the support bot, the research bot, and the operations bot each keep their own instructions, model choice, and memory.

Updated Sep 10, 20269 min read

Deployment route

01
Define the roles
Write down the two or three jobs that genuinely differ in audience, tone, or data access before creating any bot.
02
Choose a model per role
Give routine roles a fast model and reasoning-heavy roles a stronger one so cost follows value.
03
Separate memory where it matters
Decide which context is shared across bots and which should stay inside one bot to avoid leaking information between audiences.
04
Attach channels last
Connect each bot to the channel its audience already uses, and keep a test channel for changes.

Four-step path

How do you split work across multiple Hermes Agent bots?

Define the role first, then the model, then the memory boundary, and only then connect a channel.

Step 01
Define the roles
Write down the two or three jobs that genuinely differ in audience, tone, or data access before creating any bot.
Step 02
Choose a model per role
Give routine roles a fast model and reasoning-heavy roles a stronger one so cost follows value.
Step 03
Separate memory where it matters
Decide which context is shared across bots and which should stay inside one bot to avoid leaking information between audiences.
Step 04
Attach channels last
Connect each bot to the channel its audience already uses, and keep a test channel for changes.

What is Hermes Agent bot mode?

Bot mode turns a single Hermes Agent installation into a roster of named bots. Each bot has its own identity, instructions, model, memory, and channel connections while sharing the same runtime.

This is different from running several independent installations. You keep one deployment to operate, one place to update, and one runtime to monitor, while the bots stay separated by role.

When should one agent become several bots?

Split when the audiences, data access, or tone differ enough that a single set of instructions would be wrong for at least one of them.

Customer-facing support and internal operations should not share one voice.
A research bot that reads broad sources should not inherit customer data memory.
High-volume routine questions deserve a cheaper model than deep analysis work.
Community moderation and sales follow-up have different escalation rules.

Compare Hermes Agent channelsMatch each bot to the channel where its audience already is.

Should bots share memory?

Share memory only where the same context is genuinely useful to both bots. Otherwise keep it separate, because shared memory is the fastest way to leak one audience's context into another conversation.

A practical default is to share stable organisation context such as product facts, and to keep conversation history and customer specifics inside the bot that earned them.

What does bot mode change about cost and operations?

Bot mode keeps one runtime but multiplies model usage, channel connections, and review surface. The operational saving is real, but the model bill and the configuration surface still grow with each bot.

Track usage per bot rather than per deployment. When one bot dominates the bill, that is a signal to review its model choice or its trigger volume, not to abandon the split.

Estimate Hermes Agent running costsUnderstand what actually drives spend across bots and channels.

Which primary sources support this page?

Product behavior is checked against the current deployment flow. These external links provide the upstream project or channel documentation used for setup details.

Direct answers

Frequently asked questions

Can Hermes Agent run more than one bot at a time?

Yes. Hermes Agent supports running multiple named bots from one installation, each with its own role, model, memory, and connected channels.

Do multiple bots need multiple servers?

No. Bot mode is designed to share a runtime, so the bots are separated logically rather than by infrastructure.

Can different bots use different models?

Yes, and they usually should. Pairing a fast model with routine roles and a stronger model with analysis roles keeps cost aligned with value.

How many bots are too many?

When you cannot explain in one sentence what each bot is for, or when nobody reviews its output, that bot has stopped earning its place.

Key takeaways

One assistant doing every job becomes vague fast. Hermes Agent bot mode lets you split work across named bots, so the support bot, the research bot, and the operations bot each keep their own instructions, model choice, and memory.
Start with one model, one channel, and one controlled test conversation.
Use managed hosting when deployment speed matters more than operating the server yourself.

Continue the setup

Use case

Hermes Agent Team Ops Automation

Deploy Hermes Agent for team operations automation, recurring reminders, status updates, handoffs, and workflow follow-up across messaging channels.

Read next

Use case

Hermes Agent Community Assistant for Discord and Telegram

Deploy Hermes Agent as a managed community assistant for Discord and Telegram question triage, announcement drafts, moderation routing, and member support.

Read next

Blog guide