
Founding developer @flowith. Building Aha Labs — independent experiments in AI products.
Joined April 2013
- Tweets275
- Following427
- Followers31
- Likes344
introducing flowith 3.0. the ultimate canvas for human-ai creation.
two years ago, we were the first to move ai out of the chatbox and onto an infinite canvas. today, we are pushing that boundary much further.
everyone talks about full autonomy, but for real creation, keeping the human in the loop is everything. the real frontier isn't just faster agents—it's making human imagination explode.
in flowith 3.0, you orchestrate every SOTA model in one place—from Seedance to Claude Code, from GPT Image to Gemini 4 Agon.
we built native pro tools for video and image creators, and deeply upgraded text and code. you can now run cloud-native claude code and codex right on the canvas, feed them your visual assets, and ship finished work end-to-end.
we also made the pricing absurdly good:
- up to 80% off Claude Opus 5.5 & GPT Astra
- up to 95% off Seedance, Seedream, GPT Image & Muse
available today. can't wait to see what you build.
RT + comment to win a free credits
你们也在飞书里追问「这个需求做到哪了」,再找人翻 GitHub 吗?
我想找 3 个同时用飞书和 GitHub 的团队,拿一个真实问题试试 Mochi:查项目、带回出处,再按交代更新 Issue。需要 Mac 和自己的模型账号,公开测试版。
愿意试的话,回复说说场景。图为官网示例。
mochi.host/?utm_source=x&utm…
53
Working on a moodboard this week? I'm looking for 5 people to try Galta on a real reference board: collect, keep sources, then point and ask. Chrome/Dia + Mac for AI.
UI demo; agent steps re-enacted.
galta.art/en/?utm_source=x&u…
1
2
56
团队上下文隔离是生产力的最大障碍!
从此时此刻的模型能力出发,3 个志同道合、相互信任、能力互补的伙伴,如果把自己的设备和权限统统“共产”到一个平台,Agent 能获取任意角落的上下文,人类也能互访彼此的 session,彻底破除隔离,就能解放出无敌的生产力🫡
38
I put codex agent on a canvas and let it loose.
Drop in references. Scribble an idea. Point at something: “What if we tried this?”
You don’t have to figure it all out before you start.
Galta. Built by me, open source.
galta.art
1
3
103
First time sharing Galta on X. Been building this one in secret lol.
It started with a feeling I couldn’t shake: SOTA models like Opus and Astra shouldn’t be trapped in a text box. They belong on a canvas as “spatial world models.”
Just for fun, I dragged a suit-cat meme from Pinterest onto the canvas and asked the agent to draw a Miyazaki-style version using the pen tool.
Out came 620 editable vector strokes.
That’s when I got greedy.
“Again. One continuous line. Never lift the pen.”
The whole cat, suit and all, came back as a single stroke. 70,722 points without a single break—and it looked way better than I’d expected.
This started as a bit of fun. Now it’s got me wondering how far these models could go with a canvas.
1
1
182
Koho Zheng retweeted
Matrix 1.0.7 is here.
We gave it one sentence: make a launch teaser for a drink brand.
It split the work across 4 departments, picked the right models, generated keyframes + video, sent its own cut back twice, then filed everything in Notion.
New in 1.0.7:
→ Redesigned workspace
→ One model picker
→ Image & video built in
→ MCP
→ Department integrations
I just reserved my Cloudflare Wallet tag: koho.cloudflare.pay.
Reserve yours now at cloudflare.pay
76
We spend our lives staring into the light, waiting for it to show us something. Sometimes the empty wall is the whole point.
1
2
74
We've been quiet for 8 months.
Because we've been busy building the infrastructure for a 100% agent-led companies.
Still in the beta phase, but I can't hold back this preview.
Introducing Matrix, where anyone can launch a 0-Person Company that actually earns.
And yes, Matrix beta already achieved SOTA on the frontier harness, matching Fable's performance.
another stunning product 🪩
We've been quiet for 8 months.
Because we've been busy building the infrastructure for a 100% agent-led companies.
Still in the beta phase, but I can't hold back this preview.
Introducing Matrix, where anyone can launch a 0-Person Company that actually earns.
And yes, Matrix beta already achieved SOTA on the frontier harness, matching Fable's performance.
1
53








