Building https://xitter.cf/t.co/BeQDkLODLD, the coding agent for researchers MSc. in AI+Health @mcgillu

Québec, Canada
Joined August 2024
Canada’s top 10 universities right now (Official QS Rankings) 🎓🇨🇦 🏛️ The Top 10 1. McGill University (QC) 2. University of Toronto (ON) 3. University of British Columbia (BC) 4. University of Alberta (AB) 5. University of Waterloo (ON) 6. Western University (ON) 7. Université de Montréal (QC) 8. McMaster University (ON) 9. Queen's University (ON) 10. University of Ottawa (ON) UofT grads are currently typing… 💬
50
59
21
694
113,576
All your software factory complexity will get reduced to a single SKILL.md file in the future.
1
27
Ben Affleck uses dalliance.io btw
Ben Affleck reveals he writes Python, understands convolutional neural networks, worked extensively with GPUs, and used his celebrity status to get private looks at Google and OpenAI’s video models “I’ve always been kind of into computers since I was young. Then, when film started to move from analog film to digital, I became more interested in that aspect of it. The visual-effects workflow for many years has included machine learning, so I can write pretty shitty Python scripts and stuff like that. “With convolutional neural networks, which were the precursors to what the transformer can do, which is much more computation simultaneously, you would do things like look at what’s called a tensor. That’s the numerical translation of a visual image in numbers, like the batch number, the frame number and the red, green and blue values of each pixel in each frame. It’s just that simple. That numeric is called a tensor. “You’d use a convolutional neural network to identify patterns that reveal what’s called edge detection or feature extraction, which is identifying patterns well enough to know, this is where the window ledge is, so we can more easily take the green-screen image out and replace it with something. “That was familiar to me early on because, prior to Artists Equity, I had a small visual-effects company. I’ve worked with GPUs a lot too. The visual-effects guys said, ‘Hey, you should see. There are a couple: Google and this other company, OpenAI, are doing really interesting stuff with transformers in video.’ “I’ve learned that I can actually just call up and go, ‘Hey, it’s Ben Affleck. Can I come see what you’re doing?’ Sometimes people say yes, to my astonishment.”
43
I did the same on my honeymoon. Wife wasn’t happy.
on our honeymoon btw
2
37
I took a *hard* feature from a client brief to a demo video in 100 mins. - Ticket to PR (including QA): ~ 60 mins. - Demo video using Opus 5.5 + Remotion: ~ 40 mins In agents, we trust.
2
25
Give an AI agent a dataset and tell it to "keep trying until you find something significant." You've just automated p-hacking. I'm building tools for AI-assisted research, and this is the instruction I'm worried about.
1
2
39
Every Dalliance project has a brief: the goal, the data, and the rules of your study. The agent reads it before every cell. Under the hood it's a plain AGENTS.md file in your notebook's folder, the same format Codex, Cursor and other coding agents read.
2
39
Spent days on one detail: the agent has to show its assumptions before it shows its answer.
1
39
This video by @theo gave me superpowers. I genuinely can’t believe he is giving away all that for free. I revived an old MBA with Omarchy+Tailscale installed and I’m flying using it as a remote machine. youtu.be/D8PikZ1KhUo?si=cmNO…
4
1
3
128
65,095
Shipped a new landing page based on @edo_lunardi's feedback. Now you can play around with an interactive version of Dalliance right in the browser. dalliance.io
Replying to @amirominium
Looks clean and straight to the point, but also the same as other hundreds of saas websites. I would focus on details and interactions e.g. I think users love to play around with an interactive version of the product, rather than just looking at static screenshots. This alone will add some more character to it.
1
3
336
I showed Dalliance to three potential customers. They loved it. Time to push
3
256
To get something you never had. You have to do something you never did.
2
31
The scary part of using AI for research isn't bad code. It's that it quietly guesses what your data means and hands you a confident answer. So I built Dalliance. When your data is ambiguous, it asks. Every assumption is underlined and editable. The Python is one click away.
1
4
241
You didn't get a PhD to write code. You got it to find answers. Introducing Dalliance, the coding agent harness for researchers. The Jupyter notebook you know, but every cell is plain English. Ask. It writes the code and runs it. You keep experimenting. Try it at dalliance.io for FREE (bring your own key)
1
1
4
151
What makes it different: ◇ It asks when your data is ambiguous instead of guessing ◇ Its assumptions are underlined, and you can click to correct them ◇ Unchanged cells rerun with no model call ◇ Every working version of a cell is kept ◇ Notebooks save as plain .ipynb
1
1
15
Mac only for now, Apple Silicon and Intel. First launch sets up your model key and a private Python environment with pandas, NumPy and matplotlib. Claude Opus 5 by default, or any model on the Vercel AI Gateway. dalliance.io
1
12