Cheers I gave a talk on various aspects on UKSTUG few weeks ago:
https://www.youtube.com/watch?v=t8oy-5AAjqg Kasper Østerbye - AI Inside Pharo - 28 January 2026 youtube.com It is long but you can pres and check different aspects. But in my experience you need to have access to one of the better AI models. So depending on what you want, the choice actually matters (if you want to use TOOLS it turns out that Grok is the one to use at the moment). I can explain in details if any wants me to. You can download my experimental code on: https://github.com/kasperosterbye/PharoAIActions kasperosterbye/PharoAIActions github.com I have only run in on my Mac, but I am not really using any deep level stuff. There is a load mechanism there, and I you should be able to try a few Mistral models. After you have loaded it, you should be able to write in Playground: 'What is the capital of Denmark' q0. "Press command-I to see the result" I am currently experimenting with actually getting the AI to write Pharo code (and the AI compiles it into Pharo, you do not need to paste the code). Best, Kasper > On 19 Feb 2026, at 05.58, S Krish via Pharo-users > <[email protected]> wrote: > > > Looks like an opportunity to ride the Gen AI wave if there is concerted > effort to show a 10% improvement over any other language using Pharo. > > Some of the historical issues of smalltalk adoption and the key fact that > it's USP, Syntax and grammar simplicity can play on its favor... > > Will do a deep dive sometime ahead I can spare > > https://share.google/aimode/yOBigzqW18bl8nbeb > > > > On Thu, 19 Feb, 2026, 10:22 S Krish, <[email protected] > <mailto:[email protected]>> wrote: >> If we can integrate the typical Speckit or Openspec like workflow into >> smalltalk it can be an experience better than java and python too.. >> >> Maybe have a vscode folder talk to pharo through a tcp / udp pipe.. run >> through the workflow after init: >> Read the application package if brown field to baseline, then go through >> new change, apply and archive lifestyle >> >> OR >> >> have all of it in the IDE.. including md text etc .. use pharo chat to >> interact.. >> >> >> Integrating AI into Pharo Smalltalk enables a unique "Live Spec" workflow >> that surpasses the Java model. Because Pharo is a live image rather than a >> collection of static files, the LLM can interact directly with the running >> environment's Abstract Syntax Tree (AST) and reflective capabilities >> >> >> On Tue, 10 Feb, 2026, 22:08 Koen De Hondt, <[email protected] >> <mailto:[email protected]>> wrote: >>> Dear Pharo users and developers, >>> >>> I posted a second article in the series about describing Spec application >>> behaviour with Hera. The first post explained how Hera manages open Spec >>> windows. The new post explains how Hera identifies presenters so they can >>> be referenced in Gherkin scenarios. >>> >>> https://all-objects-all-the-time.st/#/blog/posts/20 >>> >>> As always, all feedback is welcome. >>> >>> Happy reading! >>> >>> Ciao, >>> Koen
