AI News Today | Julian Goldie Podcast · Julian Goldie

Hermes Agent VS OpenClaw

·14 min

The show treats Hermes agent like a direct challenge to the current agent leaders. The host says it comes from Noose Research and keeps coming back to the same point: it remembers what you do and improves the more you use it. That memory angle is the main hook. Most agents wipe the slate clean when you close them, so he keeps contrasting that throwaway workflow with a tool that carries context forward. The incumbent is the benchmark. He presents it as the big open source name in the room, the project everybody knows, but says Hermes is now making people look twice. Momentum matters here. The show points to OpenRouter personal agent categories and says Hermes is catching up because people are using it more and because it feels more reliable with smaller models. Security comes up too. The host says Hermes seems to have fewer security risks than the incumbent, which makes the switch feel less experimental. Migration is where the pitch turns practical. He highlights a one-click command that moves existing data over while preserving memories, skills, and API keys, so the switch is framed as a short job instead of a messy rebuild. That is a real selling point. The rough edges around the incumbent get a lot of airtime. He calls out crashes on model switches, unreliable tool calls, and a bloated TypeScript code base as reasons people are looking elsewhere. One developer's line about Docker being enough to sell them gets used as proof that convenience matters. The founder activity also gets attention. Technium is described as visibly engaged, answering feedback, folding in community contributions, and even turning a community-surfaced browser use integration into an official Hermes browser backend. Then the tool list lands hard, with more than 40 built-in tools like web search, browser automation, vision, image generation, text to speech, and code execution. The tone stays urgent all the way through. By the end, the episode makes one simple point: the gap is still wide, but if Hermes keeps this pace, waiting too long could mean missing the window.

As heard by us

A sharp case for memory, migration, and control in the agent race.

Hermes Agent VS OpenClaw argues that memory changes what an agent feels like to use. The host frames Hermes as something that keeps learning from a user's work, remembers preferences, and resumes instead of making people start over each time.

Read the full review in PlayNext →

Why you'd press play

A fast read on whether Hermes is the agent to try while it is still early.

Read the full recommendation in PlayNext →

Listen to the show on

Apple Podcasts Spotify