HN user

jamesfisher

2,777 karma

https://jameshfisher.com

Building https://tigyog.app

Posts81
Comments192
View on HN
www.granola.ai 18d ago

Dangerously-skip-permissions is the only safe mode

jamesfisher
4pts0
github.com 3mo ago

Tell HN: If your agent can create a PR, it can merge it too

jamesfisher
1pts0
jameshfisher.com 1y ago

Sticky snap: a better snapping algorithm

jamesfisher
3pts0
www.granola.ai 1y ago

Start with a dot-plot

jamesfisher
1pts0
www.granola.ai 1y ago

How I fixed an expensive CSS animation

jamesfisher
3pts0
www.bbc.co.uk 1y ago

Gazans tell BBC of torture in Israeli detention

jamesfisher
102pts24
www.granola.ai 1y ago

Don't Animate Height

jamesfisher
5pts1
www.youtube.com 1y ago

Your head is on backwards

jamesfisher
3pts0
www.granola.so 2y ago

How to evolve a product

jamesfisher
7pts0
slate.com 2y ago

The Poop Broker

jamesfisher
3pts0
www.youtube.com 2y ago

Statistical Rethinking: a course on Bayesian data analysis [video]

jamesfisher
14pts0
jameshfisher.com 2y ago

Shape typing in Python

jamesfisher
2pts0
jameshfisher.com 2y ago

Automatic differentiation with dual numbers

jamesfisher
2pts1
jameshfisher.com 2y ago

Tell the LLM the business context

jamesfisher
2pts0
jameshfisher.com 2y ago

It might not need a label

jamesfisher
4pts1
jameshfisher.com 2y ago

A formula for responsive font-size

jamesfisher
165pts108
slite.com 2y ago

How to use language models' tacit knowledge

jamesfisher
9pts0
jameshfisher.com 2y ago

Build Norms, Not Features

jamesfisher
2pts0
github.com 3y ago

Agent Functions

jamesfisher
1pts0
www.youtube.com 3y ago

How your brain organizes information

jamesfisher
2pts0
news.ycombinator.com 3y ago

Ask HN: Why does every package+module system become a Rube Goldberg machine?

jamesfisher
164pts213
everyday-data-science.tigyog.app 3y ago

Imputation: An Interactive Tutorial

jamesfisher
2pts0
busy-beavers.tigyog.app 3y ago

Proving programs halt: an interactive tutorial

jamesfisher
2pts0
busy-beavers.tigyog.app 3y ago

Rice’s theorem – an interactive tutorial

jamesfisher
3pts0
www.bbc.co.uk 3y ago

Patients prescribed heating as part of health trial

jamesfisher
6pts5
tigyog.app 3y ago

Rice’s theorem – an interactive tutorial

jamesfisher
4pts0
tigyog.app 3y ago

Gödel’s first incompleteness theorem – an interactive tutorial

jamesfisher
404pts107
tigyog.app 3y ago

An interactive look at A/B testing

jamesfisher
2pts0
tigyog.app 3y ago

Learn Differential Equations

jamesfisher
2pts0
tigyog.app 3y ago

Regulating email providers

jamesfisher
58pts21

Granola | London, UK | On-site | https://granola.ai

Granola is an AI notepad for people in back-to-back meetings. Our macOS app takes your meeting notes and combines them with a transcript to get an awesome meeting summary. And increasingly, Granola is a new kind of knowledge-base.

We're 15 people working in lovely Shoreditch office. See our approach to early-stage product development: https://www.granola.ai/blog/how-to-evolve-a-product

Stack: TypeScript, React, Electron, AWS, various LLM tech.

We recently raised $20M and we're growing very quickly. We're funded by Spark Capital, with participation from investors AI Grant, Lightspeed, Betaworks, Firstminute Capital, and others.

Open roles:

- Product Engineers (Backend & Frontend)

- iOS Engineer

- Electron Engineer

- AI Engineer

- Founding Designer

Competitive compensation with above-market equity.

See https://granola.ai/jobs or email hiring@granola.so.

Trust doesn’t come only from the top down [...] So, we’re also enabling trusted verifiers: organizations that can directly issue blue checks.

Is this not still a top-down system, just with one level of indirection?

Something not-top-down might look more like the web-of-trust model.

Granola | London, UK | On-site | https://granola.ai

Granola is an AI notepad for people in back-to-back meetings. Our macOS app takes your meeting notes and combines them with a transcript to get an awesome meeting summary. And increasingly, Granola is a new kind of knowledge-base.

We're 15 people working in lovely Shoreditch office. See our approach to early-stage product development: https://www.granola.ai/blog/how-to-evolve-a-product

Stack: TypeScript, React, Electron, AWS, various LLM tech.

We recently raised $20M and we're growing quickly. We're funded by Spark Capital, with participation from investors AI Grant, Lightspeed, Betaworks, Firstminute Capital, and others.

Open roles:

- Product Engineers (Backend & Frontend) - iOS Engineer - Windows Engineer (contract) - AI Engineer - Founding Designer

Competitive compensation with above-market equity.

See https://granola.ai/jobs or email hiring@granola.so.

Granola | London, UK | On-site | https://granola.ai

Granola is like Apple notes, but it also transcribes your meeting to enhance your notes. We recently raised $20M and we're growing quickly.

Stack: TypeScript, React, Electron, AWS, macOS, various LLM tech.

We're eight friendly people working in a Shoreditch office. We're product-focussed generalists. See our approach to early-stage product development: https://www.granola.ai/blog/how-to-evolve-a-product

Open roles:

- Product Engineer

- AI Engineer

- macOS Software Engineer

- Founding Product Marketer

Competitive compensation with above-market equity.

See https://granola.ai/jobs or email hiring@granola.so.

Granola | London, UK | Full-time | On-site | https://granola.ai

People in back-to-back meetings use Granola to write summaries using their transcripts and raw notes. We're backed by top investors (Lightspeed, AI Grant) and growing quickly.

Stack: TypeScript, React, Electron, Node.js, AWS, native macOS APIs, various LLM tech.

Seven friendly people working in a Shoreditch office. We're product-focussed generalists. See our approach to early-stage product development: https://www.granola.ai/blog/how-to-evolve-a-product

Open roles:

- Product Engineer

- AI Engineer

- macOS Software Engineer

- Founding Product Marketer

Competitive compensation with above-market equity.

See https://granola.ai/jobs or email hiring@granola.so.

What is "big cat DNA"? Are there not multiple species of big cat? Can a DNA test not be more specific?

Edit: Answering my own question about the methods used: The story links to Robin Allaby's page [1] about his DNA test service. That page says he tests for "the Panthera, Puma and Lynx genera". An older story [2] describes Allaby's methods: 450 PCRs and ~600 sequence reactions. "The team searched for two gene targets each of deer and canid, but over 30 different cat gene targets."

[1] https://warwick.ac.uk/fac/sci/lifesci/research/archaeobotany...

[2] https://warwick.ac.uk/newsandevents/pressreleases/research_i...

Meta Quest 3 3 years ago

"Accept all cookies" just sends me back to the cookie acceptance prompt. Can't view the page content.

Chrome on Android.

Similarly awful: those Giphy gifs that "writers" add to their business reports, content marketing etc. Never funny, always distracting. Try making your writing entertaining instead.

Traditional computation closely aligns with System 2 thinking: It’s formulaic and inflexible

Hmm, "formulaic and inflexible" is exactly how I'd describe System 1, not 2. Am I misunderstanding their analogy?

This makes so much more sense than langchain! I read all the docs for langchain, but the abstractions seemed broken and it wasn't obvious what was happening at the LLM level. In contrast, I grokked the entire llm_agents API within 5 minutes.

Why do these systems use hash-based fingerprinting? Wouldn't it be "better" to have a "browserspace vector", or "browser embedding"? So that if one fingerprint tactic fails in incognito, you don't completely lose the fingerprint, you just get a slightly different vector?

This would be a good post, if only I could read any of those images on mobile. Substack, fix your damned user-scalable=0! Even clicking on the image doesn't provide any way of zooming in on it. Do they do any usability testing?

This removes '1' but not 'I'. But if you want to avoid ambiguity, you need to remove _both_ characters, not just one. Users don't know your IDs never contain a '1', so when they see 'I', they'll still think, "hmm, is that a 1?"