Hermes Agent vs Chatbot UI — Full Agent vs Chat Interface

Chatbot UI (by McKay Wrigley) is an open-source web interface for interacting with LLM APIs — primarily OpenAI's. It provides a ChatGPT-like experience but self-hosted, with features like multiple conversation management, model switching, and custom system prompts. It became popular when users wanted the ChatGPT UX without sending data to OpenAI's app. With 30,000+ GitHub stars, it has a significant user base.

Chatbot UI is, fundamentally, a frontend. It's a React web app that wraps API calls to LLM providers in a nice chat interface. It has no persistent memory beyond what's stored in your browser's local storage (or a Supabase backend). It has no tools — it just formats your messages and displays responses. There is no self-improvement, no scheduling, no messaging integration, and no agentic capability.

Hermes Agent is the other end of the spectrum: a full backend agent with tools, memory, and self-improvement. For users who want more than a pretty interface for API calls, Hermes is categorically different.

Chatbot UI was created by McKay Wrigley in early 2023, initially as a clean, self-hosted alternative to the ChatGPT web interface. The appeal was clear: same ChatGPT experience, but you control the interface, you use your own API key, and your conversations aren't saved on OpenAI's servers. The project resonated strongly — 30,000+ GitHub stars reflect the genuine demand for a privacy-respecting chat interface.

Over time, Chatbot UI added features: multiple conversations, folder organization, system prompt templates, model selection, and a Supabase backend option for cloud persistence. These improvements made it a more capable UI. But the fundamental nature didn't change: Chatbot UI is a frontend that wraps API calls. It cannot do anything an LLM wouldn't do via API — it has no tools, no memory system, no scheduling, and no autonomous operation.

The memory distinction is worth unpacking specifically. Chatbot UI can store your conversation history in Supabase — so you can recall past conversations by scrolling back. This is flat text storage, not semantic memory. If you want to find something you discussed three months ago, you scroll through conversations. Hermes's ChromaDB episodic memory provides semantic search — describe what you discussed and Hermes retrieves the relevant context even if you don't remember when it happened.

Tools represent the most fundamental capability gap. When an LLM in Chatbot UI says 'I'll check the current Bitcoin price,' it can't actually check it — it either hallucinates a value or admits it doesn't know. When Hermes says 'I'll check the current Bitcoin price,' it calls its web search tool, gets the real current price, and returns it. This is the difference between an AI that tells you what it would do and an AI that does it.

Scheduling is another dimension where the comparison is clear. Chatbot UI has no concept of running tasks while you're not looking at it. Hermes can be configured with cron schedules to run daily reports, weekly analyses, nightly health checks. These run unattended and can push results to your Telegram — Chatbot UI cannot do any of this.

The honest limitation of Hermes in this comparison: Hermes doesn't have a web UI. Its interfaces are CLI, Telegram, Discord, Slack, and WhatsApp. For users who specifically want a web browser-based interface, Chatbot UI provides something Hermes doesn't. You could run Chatbot UI pointed at a local Ollama instance for web access and Hermes for agent tasks — many users run exactly this combination.

From a maintenance perspective, Chatbot UI is a simpler system. It's a Next.js app with optional Supabase backend — well-understood web technology. Hermes is a Python-based agent runtime with more moving parts (ChromaDB, Playwright, Paramiko, aiohttp).

For teams evaluating AI tooling, Chatbot UI is appropriate when you want a privacy-respecting, self-hosted web interface for conversational AI. Hermes is appropriate when you want an agent that can take actions, build memory, and run autonomously.

If you're using Chatbot UI for your AI interactions, transitioning to Hermes means shifting from a UI layer to an agent backend. You can run Chatbot UI pointed at Hermes's API endpoint for web access while Hermes handles memory and tools on the backend.

Export your most important conversations from Chatbot UI. Review them for durable context — project backgrounds, decisions made, conventions established. Paste this into MEMORY.md. This is the starting knowledge that Hermes will build on.

Set up the Telegram or Discord gateway for mobile access. This replaces the mobile browser use of Chatbot UI with a native messaging interface.

For the first few weeks, you'll miss the web UI. The trade-off becomes visible over the following months as Hermes accumulates memory and builds skills. By month 2, the compounding value — the agent that knows your context — outweighs the interface preference for most users.