Head of Engineering @CloudThinkerio

Joined October 2021
8 Photos and videos
Henry (Duc) Bui retweeted
I wrote this ~3 months ago, and since then, 1) Memory has been more or less fully integrated with the frontier models 2) Almost all features that made OpenClaw unique as a harness has been fully absorbed by the frontier models (e.g. schedules, loops, goals, memory, etc.) 3) New, vertical killing features and capabilities are being added every other week -- All that being said, agentic engineering is still an incredibly high skill affair. It is now obvious to me that there is a gulf of know-how and tacit knowledge between those that CAN remove humans-out-of-the-loop and actually produce a working product, and the rest of the world insisting that agents are still producing "slop".
46
140
1,840
233,958
Henry (Duc) Bui retweeted

28
242
1,629
473,882
Henry (Duc) Bui retweeted
start your tool execution mid-model call with this new eager-tools middleware! tools execute as soon as their block finishes streaming, reducing end to end latency! s/o @bmd1905 for this awesome new middleware! source: github.com/cloudthinker-ai/e… docs: docs.langchain.com/oss/pytho…
1
2
16
2,129
Henry (Duc) Bui retweeted
Introducing Mirage, a unified virtual filesystem for AI agents! 6 weeks. 1.1M lines of code. We rewrote bash from the ground up so cat, grep, head, and pipes work across heterogeneous services. S3, Google Drive, Slack, Gmail, GitHub, Linear, Notion, Postgres, MongoDB, SSH, and more, all mounted side-by-side as one filesystem. Bash that AI agents already know works on every format! cat, grep, head, and wc parse .parquet, .csv, .json, .h5, even .wav! One pipe can stitch S3, Drive, GitHub, Slack, and Linear together, same Unix semantics throughout. Workspaces are versioned too. Snapshot, clone, and roll back the whole thing with one API call. A two-layer cache turns repeated reads into local lookups, so agent loops stay fast and cheap. Drop a Workspace into FastAPI, Express, or a browser app. Wire it into OpenAI Agents SDK, Vercel AI SDK, LangChain, Mastra, or Pi. Run it alongside Claude Code and Codex. Site: strukto.ai/mirage GitHub: github.com/strukto-ai/mirage #AIAgents #OpenSource #AgenticAI #Strukto #Filesystem #VFS
171
338
3,269
618,904
7/ cc @OpenAIDevs @stevenheidel, core already has an OpenAI stream adapter. Would love your read on the cleanest fit with the Agents SDK.
336
6/ cc @langchain @hwchase17, drops into create_agent as one middleware line, no changes to the Tool contract. Would love your read.
1
58
4/ Numbers: CloudThinker production: 50% median latency cut Synthetic bench (16 workloads): 1.20×–1.50× over parallel, median ~1.28× Eager wins every workload in the suite. The model never sees a different prompt or different output — only the timing changes.
25
3/ Eager tool calling fires each tool the moment the model finishes describing it — even if the rest of the response is still streaming. Tools run during generation, not after. stream max(tool) becomes max(stream, max(tool)).
23
2/ Today, even with parallel tool calling, the flow is: Model finishes generating the full response Then all the tool calls fire in parallel The tools sit idle for seconds while the model is still typing.
23
1/ Just open-sourced eager-tools — a drop-in speedup for agents. Same model. Same tools. Same outputs. The only difference: tools start running while the model is still generating its response, instead of waiting for it to finish.
24
Henry (Duc) Bui retweeted
May 8

1,089
2,255
17,507
14,181,228
Henry (Duc) Bui retweeted
Apr 28
Our approach to hiring is a little different. We value slope over credentials, judgment over polish, and real work over interviews. Here’s what the process looks like: linear.app/now/how-we-hire-a…
15
22
644
54,147
Henry (Duc) Bui retweeted

31
126
1,127
166,134
Henry (Duc) Bui retweeted
The new paradigm: dialects MCP -> CLI -> custom harness -> dialect (skill helper files)
7
24
267
51,783
Henry (Duc) Bui retweeted
Apr 23
Introducing GPT-5.5 A new class of intelligence for real work and powering agents, built to understand complex goals, use tools, check its work, and carry more tasks through to completion. It marks a new way of getting computer work done. Now available in ChatGPT and Codex.
2,485
6,881
51,531
13,112,175
Henry (Duc) Bui retweeted
10 terminal tools that make you 10x faster in 2026: 1. zoxide A smarter cd that learns your habits. Type "z proj" and it jumps to the directory you actually meant. Repo → github.com/ajeetdsouza/zoxid… 2. fzf The fuzzy finder that powers half the terminal world. Search files, processes, git branches, shell history, anything. Repo → github.com/junegunn/fzf 3. ripgrep 10x faster than grep. Respects .gitignore by default. Once you use it, you can never go back. Repo → github.com/BurntSushi/ripgre… 4. lazygit Every git command you hate, now one keypress away. Interactive rebase feels like cheating. Repo → github.com/jesseduffield/laz… 5. starship A shell prompt that shows git status, language versions, and cloud context. Works on every shell. Renders in under 10ms. Repo → github.com/starship/starship 6. atuin Replaces your shell history with a searchable SQLite database. Syncs encrypted across every machine you own. Repo → github.com/atuinsh/atuin 7. bat cat with syntax highlighting, line numbers, and git integration. Your terminal will never look the same. Repo → github.com/sharkdp/bat 8. eza A modern ls with colors, icons, and git status built in. Makes every directory readable at a glance. Repo → github.com/eza-community/eza 9. yazi A blazing fast file manager that runs in your terminal. Image previews, async I/O, vim keybindings. Repo → github.com/sxyazi/yazi 10. delta Turns git diff into something you actually want to read. Side-by-side view, syntax highlighting, line numbers. Repo → github.com/dandavison/delta
38
314
2,842
233,409
Eager Tool Calling @ @CloudThinkerio 🚀 Agents = lots of external calls → high latency Sequential ❌ Parallel ✅ But… what if both aren’t optimal? → Built Eager Tool Calling → ~50% faster in production Open-sourcing soon. Feedback welcome 🫡
1
11