@horbunovdima i've tried all the models on your list. Each has its advantages. But I chose the GROK 4.5 💪 as my main model - it's inexpensive and the quality is excellent
3/ i don’t copy these moves
i treat them like footprints
maybe it’s nothing
maybe it’s a sharp player
maybe someone is closer to the story than the timeline
the bot watches the trail
i decide if it’s worth following
1/ i got tired of seeing “smart money” only after the move was obvious
so i built a watcher for weird wallets
- fresh accounts
- tiny history
- oversized positions
- one market
- then silence
2/ every alert needs a reason
not “wallet moved money”
more like:
why did a 4-day-old account suddenly size into this exact outcome
why this market
why now👇
openai delaying the ipo actually makes sense
private markets still pay for the ai premium
public markets will ask about cash flow, compute commitments, and what this whole thing looks like as a real company
why rush into that room before you have to
Your Hermes Agent can now adopt an animated pet: a small sprite that reacts to what the agent is doing (idle, running a tool, thinking, waiting, finishing, failing) in the GUI or TUI.
You have nearly 3000 pets to choose from via the petdex gallery, or you can submit your own.
Sharing the playbook here:
[Insert link]
Made it mainly for my own workflow - if it helps you save time, even better.
If you want the link, drop a comment and I’ll reply (or you can just find it via my profile/posts). Feedback + suggestions welcome.
i'm tired of hunting AI agent info across threads and videos.
one post covers memory, orchestration, cost - but when you build a real system, it falls apart: memory drifts, agents loop, results weaken.
so I spent weeks putting everything I learned into one doc.
Practical playbook: Production AI Agents.
Inside (only what works):
- memory that holds past 10-15 steps
- multi-agent runs without chaos
- human-like X replies that perform
- Polymarket + Hyperliquid idea flow
- fewer tokens, same quality
All from my own tests.
i believe the chat interface remains vastly underrated.
i tested rich-format responses in Hermes Agent:
→ sections
→ checklists
→ tables
→ code blocks
→ quotes
→ links
it’s the same Telegram chat.
but the functionality is completely different. See for yourself👇
@witcheer love that you actually ran the ladder and showed it’s not “1.86x or bust” - the code wins make total sense for local agents. what kernel/setup did you use for the verify pass (and did you pin batch size/seqlen)?
i asked Hermes Agent to open Chrome, go to hermesbible. com, take a screenshot, and send it back
and it just worked.
- no setup mess
- no config fight
- no stolen cursor
just the new Computer Use feature in Hermes Agent doing the task in the background.