intellectronica Profile Banner
Eleanor Berger Profile
Eleanor Berger

@intellectronica

Followers
3K
Following
10K
Media
872
Statuses
9K

πŸ‘©β€πŸ’» Expert AI Leadership (ex @google @microsoft startups ...) πŸ‘©β€πŸ’Ό Consulting https://t.co/IeRoS5BR3f πŸ‘©β€πŸ« Teaching https://t.co/D5YmOH0muP

ZΓΌrich, Switzerland
Joined January 2011
Don't wanna be here? Send us removal request.
@intellectronica
Eleanor Berger
18 hours
One cool thing @julesagent has which Codex and Claude Web don't yet is an API (GitHub Copilot Coding Agent has an "undocumented" API which I use but can't recommend). There's a lot you can do when you can trigger agentic coding jobs from an event or on schedule.
1
2
5
@intellectronica
Eleanor Berger
24 days
AI agents treat instructions as one text stream β€” files don't matter to them. But for you? Split into MD files when it helps. Start with one, divide as needed.
Tweet card summary image
elite-ai-assisted-coding.dev
Q: Should I organize agent instructions across multiple markdown files or keep everything in one document?
0
2
3
@intellectronica
Eleanor Berger
17 days
Are Skills the new Apps? πŸ“² I converted my entire web app into a single, portable AI skill. No more servers, no more deployments. Just pure, portable capability.
Tweet card summary image
elite-ai-assisted-coding.dev
Converting a Simple Application into a Skill
0
3
5
@intellectronica
Eleanor Berger
1 day
Awesome quality-of-life upgrade: I used to have to choose between my favourite editor ( @code ) and the easiest way to run a notebook ( @GoogleColab ). With the new Colab extension ... I get the best of both worlds. https://t.co/jBvoDuJs6l
3
1
6
@intellectronica
Eleanor Berger
5 days
Claude Code rocks for fast implementation, but GPT-5-high crushes analysis & planning! πŸš€ Command: /codex to prompt it from Claude, get detailed plans, then build with Sonnet/Haiku.
Tweet card summary image
elite-ai-assisted-coding.dev
Claude Code with Claude Sonnet 4.5 and even Claude Haiku is fast, smart, and fun to work with.
0
2
4
@intellectronica
Eleanor Berger
3 days
AI agents that talk to each other and to you β€” no code, just live objects passing messages inplain English. IntelliMorphic: the lost dream of live programming environments, rebooted for the age of AI.
Tweet card summary image
elite-ai-assisted-coding.dev
IntelliMorphic @ AI Baguette
0
3
2
@intellectronica
Eleanor Berger
9 days
The people we follow and try to learn from can be divided into two categories: 1. People who tell us "this is more complicated than you think" 2. People who tell us "this is simpler than you think" I am only interested in (2) and I'm doing my best to be part of that category.
2
1
11
@intellectronica
Eleanor Berger
5 days
⭐⭐⭐⭐⭐ feedback from the first cohort of Elite AI-assisted Coding. @isaac_flath and I built a course that brings together the frameworks and techniques for applying AI in every aspect of software development. It worked. πŸ‘Œ πŸ‘‰ Join us for cohort 2: https://t.co/Z5XQG1cLW2
0
1
3
@intellectronica
Eleanor Berger
17 hours
0
0
2
@intellectronica
Eleanor Berger
18 hours
One cool thing @julesagent has which Codex and Claude Web don't yet is an API (GitHub Copilot Coding Agent has an "undocumented" API which I use but can't recommend). There's a lot you can do when you can trigger agentic coding jobs from an event or on schedule.
1
2
5
@isaac_flath
Isaac Flath
8 days
Faced a classic async UI bug: a "processing" pill didn't update I used AI to brainstorm to get 4 patterns (I chose self-polling pill) AI is good at giving common patterns fast so I can spend more time on decisions and considering less obvious options https://t.co/5ZgYkZUmRo
Tweet card summary image
elite-ai-assisted-coding.dev
I ran into a classic background task UI problem in my app this week.
0
1
2
@intellectronica
Eleanor Berger
5 days
Managing my co's expenses is never fun, and there are always some complications (like two invoices in one scanned PDF). @warpdotdev with Skillz ( https://t.co/hTtPj2MgYU ) and the PDF skill helped make it much less awful today πŸ™‚
0
1
2
@intellectronica
Eleanor Berger
3 days
AI toolkit update: πŸ’₯ Claude my daily driver. Skills and agency win. Still use GPT-5 for tough problems, Gemini for writing. Ditched Perplexity. Raycast & Notion all the things. VSCode + copilot, claude, codex and Warp for coding.
Tweet card summary image
everything.intellectronica.net
This is a follow-up to my March 2025 report on the AI tools I use.
1
1
12
@intellectronica
Eleanor Berger
5 days
πŸ’» How I use @raycast to drive a lot of work with AI tooling FAST! πŸš€ When working with the best AI tools, productivity improvements are often constrained by the human operator's ability to work fast. This is one solution that works for me. https://t.co/nrFoWli1V7
0
1
10
@intellectronica
Eleanor Berger
1 day
USE MORE WORDS
@AmandaAskell
Amanda Askell
9 days
I do think people often err on the side of trying to make their prompts too succinct, even if the idea they're trying to move from their own brain into the model's brain is very complex. I have some >100 page prompts that I use pretty regularly.
0
0
0
@intellectronica
Eleanor Berger
1 day
Sometimes companies are so blinded by NIH they hurt themselves not copying the best. Example: Codex / Gemini / Copilot CLI could be so much more successful it they just decided to copy Claude Code feature-by-feature compeltely, just with access to their respective models.
1
0
3
@intellectronica
Eleanor Berger
1 day
Awesome quality-of-life upgrade: I used to have to choose between my favourite editor ( @code ) and the easiest way to run a notebook ( @GoogleColab ). With the new Colab extension ... I get the best of both worlds. https://t.co/jBvoDuJs6l
3
1
6
@intellectronica
Eleanor Berger
2 days
Vibe Coding == Faster Horse AI-powered productivity and creativity are not going to be the result of everyone coding their own saas webapp clone. It will be the result of people realising that they can now just do things. With words, imagination, and agency.
0
1
1
@intellectronica
Eleanor Berger
2 days
Vibe Kanban β€” @LouisKnightWebb on managing multiple agents locally with full control. Dive into planning, iteration, and the QA revolution ahead. πŸš€
Tweet card summary image
elite-ai-assisted-coding.dev
Building the Future of Parallel AI Development
0
2
1
@itsohqay
Ohqay
2 days
@gooby_esq @intellectronica Those creative writing benchmarks are worthless. Outputs are graded by other generic AI’s that haven’t even been RLHF’d to score like expert human writers. The actual consensus among writers with experience using different models is that Claude Sonnet and Opus are by far the
0
1
3