Lightning talks Wednesday
Lightning talks on Wednesday at the Plone Conference 2026
Martin Peeters: Plone Foundation membership
If you are a Plone Foundation member, make sure you have renewed your membership. If not, talk to an admin. And you should have received an email allowing you to vote in the Foundation Board election.
Mack Palomaki: Aurora
I helped a musician, Salamon Jakobsson with a Plone site (Aurora on Nick), and in thanks and answer he wrote a piece of music called Aurora. Hear him explain about it, and then play it: https://salomonjakobsson.ljusa-tider.se
Astrid: Coffee disloyalty
In my town there is a new coffee shop each week. You wonder how they make any money. They are not dependent on early adopters, everyone already knows what coffee is.
Last year after the Plone Conference we were in Helsinki. It was during the Helsinki coffee week. At a coffee shop we got a disloyalty card. Not loyalty, but disloyalty. Go to a different coffee shop and you get another sticker. The sixth coffee is free.
This got us all over Helsinki.
Sometimes you just need to be relaxed, and confident.
David Glick: Recycle bin
I want to look at (my) software and want to see the beauty and the thought that was put into it.
But in Plone code you often see very old TODO comments from before a first release. Or a PLIP from seven years ago: Plone should have a recycle bin.
Now we have plone.recyclebin. You delete something, everything happens as normally, but a reference is kept. Lots of credit to Rohan Shaw who worked on this last summer. Various people reviewed it. I took time this summer to finish it, and last night I pushed the button.
Now go and use it production, and give feedback.
https://github.com/plone/plone-recyclebin
Franco: collective AI settings
Control panel for configuring a connection to an AI backend. Local or remote. You can add several models. See https://github.com/collective/collective-ai-settings
There is another package with tools that use this. When editing a page in Plone you can ask it to create a summary, or write a text. On an image you can ask it to suggest tags, description, etc. See https://github.com/collective/collective-ai-tools
Jens Klein: Doc style skill
I let AI write some documentation for Plone, but got feedback from Steve about things that were wrong with it. So I created an AI skill, based on the documentation that we have about writing documentation. Try it, it helps. See https://github.com/plone/plone-doc-style-skill
Torben Baumgartner: Show my setup: Pylon and split
Pylon is a VS Code MCP Server. See https://github.com/toalba/Pylon
And split is a proxy, that I cannot share, that can direct queries to other LLMs.
Armin: Social media engagement boost to save Plone's future
The LinkedIn Bell. For Plone to survive it should be recognized. How can you personally boost Plone? What we have is media assets: text, images, media. What we miss is range and attention.
Change your behavior, tiny habits, find the trigger. Click the bell on the Plone page on LinkedIn. Click it, and select Everything, instead of only the most important.
If you have a nice story, say from the Plone conference, share it. Then others who have clicked the bell, can see this too, and their network sees it.
Rikupekka: next.plone.org
We have been working on the next plone.org website this year. Check out the video. And join the sprint this weekend.
Armin: Show your setup: Digital KungFu
Automator is an app that is standard on Mac. You can create an action with it to run a Python script on hotkeys on the Mac. Or on iPhone.
Johan: The back of the badge
The conference badge half of the time has the blank side on the front. Please grab a pen and write something on the back of your badge.
Antoine: Plone 4 to Plone 6, make no mistake
You can try migration using agentic coding. Make no mistake: well, make mistakes, but make sure to not make the same mistake twice. Design feedback loops: have a goal you can check, give feedback fast, improve.
Loop 1: Write your migration into a big plan. Let AI turn it into smaller chunks.
Loop 2: Unit test loop: run the tests, they fail, fix one thing, run the tests. Small steps.
Loop 3: The design loop. Take the design, the AI agent builds it, compare, repeat. Whenever the AI made mistakes, we had it write a rule.
60 hours of Claude running, about 4 billion tokens consumed, on a single Claude 5x plan.
You still need a human to set goals, and give early feedback.
Alexander Loechel: Terminal status panel
Probably people often come to you with their problems, because you are a smart person. But we are lazy, and want to get something quickly done.
Something goed wrong on a server. What is the status? We developed something in house, published it as a tool. If we log into a server, in a separate terminal it shows the status of the server, all kinds of information. And you can just click some URLs in the the text it shows in the terminal.