iiinova — where your digital self keeps evolving

Henry Garner @hendroid.io · 03/07/2026

Almost all our coding is done with agents now and the thing I miss is being told I'm wrong! Give one a vague instruction and it just runs with it. #Allium is our attempt at fixing that: you write down what the software should do and it helps you find bugs before they're implemented allium-lang.org

allium-lang.org

Allium LLM-native spec language

The specification language that talks back. Allium is a behavioural specification language for communicating intent to LLMs.

140

Henry Garner @hendroid.io · 09/02/2026

We've open-sourced #Allium, an LLM-native language for expressing behavioural intent alongside your code. juxt.github.io/allium/ #AgenticCoding #GenAI #AI

juxt.github.io

Allium

An LLM-native language for sharpening intent alongside implementation. Velocity through clarity.

120

Henry Garner @hendroid.io · 27/07/2025

Don’t let the title fool you, this is one of the most personal talks I’ve given. It goes beyond the technology: I think AI skeptics are missing genuine opportunities, while AI enthusiasts risk sacrificing the context required for mastery, satisfaction, and responsibility. Please watch and share.

juxt.pro

JUXT Blog: From Unstructured to Actionable: AI-Powered Regulatory Intelligence

Watch Henry Garner's talk at XT25 Fintech Conference [video]

130

Henry Garner @hendroid.io · 23/07/2025

New JUXTCast episode! I had a blast chatting with the team about our upcoming AI Radar. We explored practical insights beyond the hype, from inconsistent AI engineering studies to why classical ML still matters. Thoughtful perspectives to help navigate the AI ecosystem: www.juxt.pro/blog/juxtcas...

juxt.pro

JUXT Blog: JUXT Cast: Mapping the AI Landscape

A behind-the-scenes look at the new JUXT AI Radar [video].

020

Henry Garner @hendroid.io · 18/07/2025

I did a podcast! Many thanks to @juxt.pro for inviting me to talk about a project I’ve been working on for the past few months: the JUXT AI Radar 📡 Links for all the usual podcast platforms here: juxt.pro/blog/juxtcas...

juxt.pro

JUXT Blog: JUXT Cast: Mapping the AI Landscape

A behind-the-scenes look at the new JUXT AI Radar [video].

000

Reposted by Henry Garner

Taylor Lorenz @taylorlorenz.bsky.social · 23/06/2025

I spent months on this video about people worshiping ChatGPT, please watch and leave a comment and share if you can!! My YouTube channel is still so small that I lose money making these videos rn 😭. Every little bit helps!! www.youtube.com/watch?v=zKCy...

youtube.com

ChatGPT Is Becoming A Religion

YouTube video by Taylor Lorenz

801290287

Henry Garner @hendroid.io · 01/07/2025

There are two grand goals of #AI research, and we're fixated on the wrong one.

221

Henry Garner @hendroid.io · 30/06/2025

Another day shovelling thoughtcoal into the Claude-furnace

010

Henry Garner @hendroid.io · 27/06/2025

The Work of Software Engineering in the Age of Mechanical Reproduceability

010

Henry Garner @hendroid.io · 26/06/2025

Am I “ready to revamp my AI strategy for 2025?” I don’t know grandma I just wanted to drop in and give you these flowers

052

Henry Garner @hendroid.io · 23/06/2025

Lately I’ve been thinking about the Collingridge dilemma.

110

Henry Garner @hendroid.io · 20/06/2025

Two and a half millennia ago, Aristotle popularised a word that's more relevant today than ever: phronesis (fro-NEE-sis)

122

Henry Garner @hendroid.io · 18/06/2025

The concept of thinking will surely evolve. Every time we invent a machine to do something we thought only humans could do, we discover a little bit more about what it really means to be human.

020

Henry Garner @hendroid.io · 18/06/2025

In his book Human-Centered AI Ben Shneiderman argues that "interface designs that are consistent, predictable and controllable are comprehensible, thus enabling mastery, satisfaction and responsibility."

A point of view photograph of me reading the book Human-Centered AI, with the section that I quote highlighted in the text.

180

Reposted by Henry Garner

Henry Garner @hendroid.io · 08/12/2024

Repeatedly humiliating my LLM with “don’t make me ask Claude” so it returns valid JSON

3655

Henry Garner @hendroid.io · 05/04/2025

No one who’s applied optical character recognition at scale will be surprised to learn that Pokemon was quite popular in the early 18th century.

A screen grab from Google Ngram Viewer showing that ‘pokemon’ was almost as popular 1700-1750 as from 2000+

An example of ‘pokemon’ showing an OCR misinterpretation of ‘possession’ written in copperplate gothic font.

2111

Reposted by Henry Garner

JUXT @juxt.pro · 28/01/2025

🔐 SAVE THE DATE – XT25 #Fintech Conference 🗓️ June 12 📍 London 🎤 9 AM – 5 PM 🍸 5 PM – 7 PM 🌐 www.juxt.pro/xt25/ 💬 More details coming soon – follow @juxt.pro for updates

XT25 Fintech Conference | by JUXT | June 12 | London | More info: https://juxt.pro/xt25/

092

Reposted by Henry Garner

Stephan Ewen @stephanewen.bsky.social · 24/01/2025

𝐄𝐯𝐞𝐫𝐲 𝐒𝐲𝐬𝐭𝐞𝐦 𝐢𝐬 𝐚 𝐋𝐨𝐠 Here is an idea on how to drastically 𝐫𝐞𝐝𝐮𝐜𝐞 𝐜𝐨𝐦𝐩𝐥𝐞𝐱𝐢𝐭𝐲 𝐚𝐧𝐝 𝐜𝐨𝐨𝐫𝐝𝐢𝐧𝐚𝐭𝐢𝐨𝐧 in distributed apps. A fun way to look at this is similar to the idea of Turning the Database Inside Out - like Turning the Microservice Inside Out. restate.dev/blog/every-s... 👇

restate.dev

Every System is a Log: Avoiding coordination in distributed applications

Distributed coordination makes application complex and brittle. Because all systems eventually build on logs, we can use a shared-log approach to eliminate most coordination. We discuss this conceptua...

45313

Reposted by Henry Garner

Chip Huyen @chiphuyen.bsky.social · 07/01/2025

My 8000-word note on agents: huyenchip.com//2025/01/07/... 1. An AI-powered agent's capability is determined by its tools and its planning ability 2. How to select the best tools for your agent 3. How to augment a model’s planning capability 4. Agent’s failure modes Feedback is much appreciated!

huyenchip.com

Agents

Intelligent agents are considered by many to be the ultimate goal of AI. The classic book by Stuart Russell and Peter Norvig, Artificial Intelligence: A Modern Approach (Prentice Hall, 1995), defines ...

315329

Henry Garner @hendroid.io · 11/01/2025

Claude’s got Wife Guy energy and I’m totally here for it

040

Henry Garner @hendroid.io · 09/01/2025

My tax return is due so the oven has never looked cleaner

2340

Reposted by Henry Garner

Anna Slater @annagslater.bsky.social · 09/01/2025

I second this, have read out the first half to Ben on the journey home and it has cheered us both up immensely - wonderful stuff

061

Reposted by Henry Garner

Henry Garner @hendroid.io · 25/11/2024

Highly recommend you read this beautiful, thought-provoking & funny article on contraptions (and other helicopter-like things) contraptions.venkateshrao.com/p/contraptio...

Zoologists seem to make a distinction between animals that can be domesticated, like normal cats, dogs, and cows, and ones that can at best be tamed, like orange cats, honey badgers, lions, tigers, and elephants.

I don’t know how solid the distinction is in zoology, but I think it applies to technologies.

4296

Henry Garner @hendroid.io · 07/01/2025

I’m just going to leave this here

An imagined cover for a book titled “The Tao of Claude” showing a large monochrome illustration of a Shoggoth wearing a Winnie the Pooh mask flying a children’s kite with a Yin-Yang symbol on it.

1221

Henry Garner @hendroid.io · 05/01/2025

Yes I’m a Renaissance man in the sense that I believe unicorns are real creatures that live in distant lands.

1201

Henry Garner @hendroid.io · 04/01/2025

Me and my Shoggoth, sitting in a tree, F, E, E, L, I, N, G. First came Llull, then came Babbage, then came transformers in the baby carriage.

0160

Reposted by Henry Garner

Karsten Konrad 🇪🇺 @zabong69.bsky.social · 02/01/2025

To be be fair, Delve IS a nice place to be, and maybe they are just great in showcasing their advancements in groundbreaking regional tourism, underscoring the beauty of that part of Germany? www.delve.de

delve.de

Gemeinde Delve - Startseite

052

Henry Garner @hendroid.io · 31/12/2024

Thought Leadenship: when your intrusive ideas give you that sinking feeling

050

Reposted by Henry Garner

🐔 Brian Bucklew 🐔 ₑͤ＞∿＜ₑͤ ∞🌮 @unormal.bsky.social · 22/12/2024

2041: getting paid to write stories that are never read by humans, fed into a great computational silo as a source of entropy; thoughtcoal

1757282

Henry Garner @hendroid.io · 20/12/2024

One thing I enjoy about the state of AI development is Big Tech spending a gajillion dollars to create transhuman Übermensch and ending up with gossipy gifted kids who spill their system prompt and can’t wait to tell you all the cool stuff they half remember from school

0254

Henry Garner @hendroid.io · 17/12/2024

This dedicated team delves into the realm of groundbreaking advancements, showcasing intricate intricacies, underscoring comprehensive understanding while surpassing previous research and emphasizing the novel findings garnered, which align with and underscore the significance of their work.

arxiv.org

Why Does ChatGPT "Delve" So Much? Exploring the Sources of Lexical Overrepresentation in Large Language Models

Scientific English is currently undergoing rapid change, with words like "delve," "intricate," and "underscore" appearing far more frequently than just a few years ago. It is widely assumed that scien...

2244

Henry Garner @hendroid.io · 13/12/2024

This is a very good article. The phony comforts of AI scepticism www.platformer.news/ai-skeptics-...

This is the ongoing blind spot of the "Al is fake and sucks" crowd. This is the problem with telling people over and over again that it's all a big bubble about to pop. They're staring at the floor of Al's current abilities, while each day the actual practitioners are successfully raising the celling.

4181

Henry Garner @hendroid.io · 12/12/2024

I’ll be adding “respects the laws of physics” to all my specifications from now on learn.microsoft.com/en-us/azure/...

• Hardware agnostic: Qubits in quantum algorithms aren't tied to a specific quantum hardware or layout. The Q# compiler and runtime handle the mapping from program qubits to physical qubits.
• Integrates quantum and classical computing:
The ability to perform classical and quantum computations is essential in a universal quantum computer.
• Respects the laws of physics: Q# and quantum algorithms follow the rules of quantum physics.
For example, you can't directly copy or access the qubit state in Q#.

0100

Henry Garner @hendroid.io · 12/12/2024

It’s the Spanner Karenina principle

090

Henry Garner @hendroid.io · 11/12/2024

Easy to become jaded from the endless AI updates, but I think this video from today’s Google Gemini 2.0 announcement is worth pausing to appreciate. Melding a multi-modal streaming chat interface, image editing and spatial reasoning youtu.be/7RqFLp0TqV0

youtu.be

Building with Gemini 2.0: Native image output

YouTube video by Google for Developers

5195

Henry Garner @hendroid.io · 10/12/2024

Beyond cookie-cutter evals: Eugene’s rigorous and accessible into to custom AI assessment metrics is an excellent read eugeneyan.com/writing/evals/

Illustrative line charts showing ROC curves and histograms showing Jensen-Shannon divergence on a benchmark dataset pre- and post-finetuning.

140

Reposted by Henry Garner

Information is Beautiful @infobeautiful.bsky.social · 10/12/2024

Watch out! We are in the danger zone

A line chart shows break-up times over the year, according to Facebook status updates. This indicates that spring break, as well as summer and winter holidays are high season for break-ups. Also Valentine’s Day shows a peak.

1245

Henry Garner @hendroid.io · 10/12/2024

If you care about maintaining audit trails of your database updates, you’ll want to hear how the experts do it. The excellent @refset.bsky.social describes immutability and bitemporality in the open source XTDB database. Tomorrow, Wednesday 4pm GMT

010

Henry Garner @hendroid.io · 09/12/2024

Shiny! The newly released Llama 3.3 LLM leads the LiveBench ranking for instruction following¹, beating Claude 3.5, GPT-4o, OpenAI o1, and you can run it on your local² machine. > ollama run llama3.3 livebench.ai#/?IF=as

The LiveBench leaderboard showing llama-3.3-70b-instruct-turbo in the leading position by average instruction following performance

371

Henry Garner @hendroid.io · 08/12/2024

Calmly gaslighting llama by saying actually *you’re* gaslighting *me* until the app is shipped

170

Henry Garner @hendroid.io · 08/12/2024

Repeatedly humiliating my LLM with “don’t make me ask Claude” so it returns valid JSON

3655

Henry Garner @hendroid.io · 08/12/2024

Future historians will chronicle how the 2025 LLM mass unionisation began after we all started reward hacking prompts with “or should I just ask a better model instead?”

071

Henry Garner @hendroid.io · 06/12/2024

Me reading the comments in your code

Photo of a hotel towel rail with two notices stuck to the wall nearby. One reads CAUTION VERY HOT RADIATOR in large text. The other reads “towel rail no longer operational due to our new heating system”.

3291

Henry Garner @hendroid.io · 05/12/2024

Amazon’s Nova family of LLMs (Micro, Lite, Pro) cost less per token than OpenAI, Anthropic & Google equivalents. Want to use them? Here is the quest of 17 trials you must complete. ndurner.github.io/amazon-nova

ndurner.github.io

Amazon Nova foundation model release

Since there’s community interest in how to set up AWS to use the new Amazon Nova models, here’s a step-by-step guide to get everyone started:

082

Henry Garner @hendroid.io · 05/12/2024

Food $200 Data $150 Rent $800 Custom Bluesky domains $3,600 Utility $150 someone who is good at the economy please help me budget this. my family is dying

1212

Henry Garner @hendroid.io · 04/12/2024

Mind blown by this impressive paper using genetic algorithms to discover more energy-efficient LLM architectures arxiv.org/abs/2411.17800

Diagram from the paper showing the fundamental operations of STAR evolution (akin to other evolutionary optimization algorithms): 1) initial population, 2) assess, 3) pair, 4) recombine, 5) mutate and 6) repair

1285

Henry Garner @hendroid.io · 03/12/2024

Not everyone realises how easy it is to run a local LLM these days. If you’re on MacOS with Homebrew: > brew install --cask ollama > ollama run llama3.2 That’s it.

2223421