🇪🇺🇦🇹 Tech-Enthusiast. AI Tinkerer. Into @OpenClaw. Former Tech Editor. Prev Apple PR; Burson - Qualcomm, Google, Accenture. Supporter of 🇺🇦.
Munich, Bavaria
Joined July 2008
- Tweets1.2K
- Following779
- Followers5.9K
- Likes3.8K
Ben Miller retweeted
The first AI persona you’ll actually want to meet face-to-face.
Most AI personas stare blankly while you talk. Emma reacts while you’re still speaking.
After 18 months of research at @nuance_ai, this is Emma, built on Nuplex, our full-duplex audiovisual foundation model.
Ben Miller retweeted
OpenCode uses 41% less memory with Bun AoT and OpenClaw starts 3x faster
ALT Bar charts titled "Compiled ahead of time with Bun", comparing Bun AoT (bun compile) with Bun's JIT and Node v26. Lower is better. OpenClaw startup, gateway ready: AoT 2.75 s, Bun 6.66 s, Node 8.62 s. 3.1× faster than Node. OpenClaw memory, RSS when ready: AoT 732 MB, Bun 882 MB, Node 1,181 MB. 38% less than Node. opencode startup, first frame: AoT 0.46 s, Bun 0.65 s. 1.4× faster. opencode memory, peak over 20 turns: AoT 421 MB, Bun 712 MB. 41% less. Linux x64, 12 cores, medians of 10+ runs, each project's own build script, unchanged. For opencode 2.1.4, the Bun bar is today's npm release. Experimental: oven-sh/bun#48120.
so lifelike, little dude even imitates the nervous head tremor and the twitching facial expressions of a typical FDE
Tap its head and it changes mood: new eyes, new moves 🤖
@DiyBurner's Desktop Companion V2 runs on our XIAO ESP32-S3, with OLED eyes, touch and a servo, and firmware built with our CodeCraft AI assistant.
Check it out on @instructables 👉 tinyurl.com/yntu6ytr
2
142
It's fascinating and a bit amusing to me that agents or LLMs in general have no (accurate) sense of time.
🧔♂️ "How long will this probably take?"
🤖: "About 4 days."
🤖, 20 mins later: "done."
Of course agents can know what time it is and their training data obviously also contains all kinds of mathematical papers and philosopcial writings about the concept of time, but they cannot assess themselves and their actions within this human framework. They have a really hard time (pun intended) estimating time and timespans. They get the mathematical aspect of time, but not the subjective one.
There are all kinds of serious and quriky benchmarks out there and I wonder if besides drawing a pelican riding a bike, mastering/estimating time should be one as well. "How long will this 10 step coding project take?" and the better the initial estimate, the higher the score.
3
3
302
Ben Miller retweeted
A new model shouldn't mean waiting for a new OpenClaw release
@obviyus goes inside his work on model catalogs, account-based discovery, and Codex versioning
From our engineering series: openclaw.ai/blog/use-new-mod…
🤖 Made with AI
Ben Miller retweeted
Engineering, Product, Marketing, each in its own silo, with its own context
OpenClaw Multiplayer gives teams one shared agent and shared context, so they can actually work together
@Pishgar on what we showed at GitHub Universe:
openclaw.ai/blog/shattering-…
Ben Miller retweeted
I’m humbled by my product and my team 🫶. Major keynote and 100% built on @openclaw with voice notes on a train from Hangzhou to Shanghai.
Never touched Google Drive. Claw had full access to my X, past talks etc. Even the self roasting pablo meme it generated i never asked 👌
Ben Miller retweeted
NEW: Apple is cutting employee AI spending budgets even as it pushes workers to use the technology more. Engineers have seen daily limits cut from $300 to $200, with further reductions to $100 planned. Other teams could see limits fall to $25. bloomberg.com/news/articles/…
Ben Miller retweeted
Knowledge: A New Approach to Memory in OpenClaw | The ClawCast Episode 16 teapaw.org/i/broadcasts/2gqWlBaea…
Ben Miller retweeted
When our team first moved to multiplayer sessions, where everyone can read each others prompts, I was initially very hesitant
I say a LOT of dumb things to my agent
That is still true, but I've learned two things since gaining full visibility into all the sessions of my teammates:
1. Everybody else says dumb things too
2. There is a ton to learn from how other people on the team drive their agents
For example, I wanted to learn how our UI/UX guy, @vhbrzezowski , prototypes with agents - so I just asked!
My agent read all of his sessions and helped me understand Vyctor's workflow using our `show_widget` tool to iterate on mockups built from our actual frontend code
Once you get over the fear of sharing your prompts with others it tears down so many useless knowledge silos across your whole company.
Ben Miller retweeted
OpenClaw v2026.10.10 🦞
🗂️ Chat Details pulls up beside wide chats
🧭 Advisor suggests course corrections
🎙️ Browser Talk gets split speech
3,028 source PRs · 454 contributors.
There’s more in the full notes docs.openclaw.ai/releases/20…
