Want a stunningly good LLM service for only $USD 10 a month?

ChatLLM by Abacus.AI

It's unreal.

If you use NGINX, you might want to take a look at Caddy

If you need a reverse proxy that is simple to set up and ‘just works', check out Caddy

I think you'll like it. 

Behind the scenes of an AI request

 Yesterday, we posted a video demonstrating our new integration with Verse.

https://pecdm3.preemptive.com.au/videos/verse-demo-1.mp4

In the video, we demonstrate the process of proofreading, translating, and asking questions.

When Preemptive AI for Domino is ‘asked to do something" it performs the following steps:

- Creates a AI-Request document

- The AI-Proxy database then processes the request. It, in turn, performs the following steps:

- Looks up the instruction type for the request, for example, AI-Proofread. This contains the prompt that is sent to the LLM; for example, our default prompt for proofreading is:


- looks up details for the service the Instruction should be sent to, e.g., Domino IQ, LM Studio, OpenAI, etc

- Sends the request

- The system processes the response and sends it back to the request document.

Here is a full example of the Demo’s request for proofreading:

This is a ripper - We’re bringing AI to HCL Verse!

We are absolutely stoked to announce that we're finally bringing our Preemptive AI for Domino functionality to HCL Verse on-premises.

Why? Well, let’s be honest, with all that has happened in the world of AI, Verse is currently sitting on the sideline like a cheerleader with a bunch of busted pom-poms.

In this upcoming release, we’re packing in the "Big Four": Proofread, Enhance, Translate, and Ask. It’s all powered by Preemptive AI for Domino and plays nice with whatever you’re running—Domino IQ, Ollama, LM Studio, OpenAI, local or remote. You name it, it’ll probably work.

So how does it work?

We did try to go down the official Verse API route, but yeah, nah, no can do. It was just too limiting for what we wanted to achieve. So, we took the "scenic route" and built it via browser extensions for Chrome, Edge, and Firefox. Sorry to the Safari users out there, not for now (send money and it can happen!). 

The Firefox version is already done and dusted, and now we’re just cooling our heels waiting for Google to do its thing and send back our signed package... fingers crossed it lands any day now. While we’re waiting for the big wigs to sign off, we’re busy stress-testing the life out of it. 

Here is a video of it in action --> https://pecdm3.preemptive.com.au/videos/verse-demo-1.mp4

While we complete final testing, we’d love for a few legends to give it a burl and provide some feedback before we ship it for real

We have set up a sandboxed environment that will do all the processing; you use your Verse client, and our server will do the AI work for you. The AI Server component for this Demo is a Mac Mini M4 Pro, running LM Studio, with the OpenAI GPT-OSS 20 billion model (mlx) loaded.

There is no easier way to try it out. 

If you’re keen to have a go, create a ticket at https://preemptive.freshdesk.com, let use know what browser your are using and we’ll sort out the details.

When it is all ready, we will make an offical announcement and make it available to everyone. 

Until then, stay frosty.

Summer holidays done and dusted

It was a pretty hot start to Summer holidays 2025 with 43C (110 F) —so camping wasn’t ideal. 

But once we got everything set up, it was all good. We had a couple of wet days, but we still managed to hit the beach or play pickleball almost every day. 

The locals were friendly.  

and there was plenty of time to relax and think,  well, about not much at all… a fantastic reset. 

Already looking forward to the next time we get away. 

Season's Greetings from Downunder!

Wishing you and your loved ones a Happy and Safe Christmas and a wonderful New Year!!! 

Have a good one !

Image by Richard Galbraith https://www.dustydog.com.au 

RAG and Domino: A prototype we built while we wait for Domino 2026

As the year comes to an end, I wanted to share a quick update on a side project we’ve been working on.

Regular readers will know that we have spent a considerable amount of time developing Preemptive AI for Domino, which is an end-to-end AI solution for Domino versions 12, 14, and 14.5. Along the way, we even added audio-to-text transcription. If you missed that series, you can find some the details here

One of the biggest practical challenges with LLMs is that to enable a model to understand your data, you must provide it securely, in a useful format, and in a way that scales.

There are several ways to do that, and one of the most common is Retrieval-Augmented Generation (RAG).

So to learn more, we built a prototype.

The goal was to build a system where we could query a knowledge store (Domino) and have an LLM respond using the most relevant source material from that store.  We used three months of my email for this experiment, which made it easy to validate the results.

At a high level, the app worked like this:

1. Extract text from emails, clean it, and store it in corresponding JSON files.

• The text was split into chunks suitable for embedding (about 1,500 characters each).
• Words are not split across chunk boundaries.
• Chunks include a 5-word overlap to preserve context.

2. Generate embeddings using the nomic-embed-text model via a local Ollama server.

• Each chunk of text got its embeddings and was turned into a vector in a matrix that has 768 dimensions. The embeddings capture meanings, not so much keywords. The math here is unbelievable and a big part of the magic that makes this work. 
• Embeddings are stored in a local vector-enabled database (not Domino).

When it is time to answer a question, we follow these steps:

3. Run queries against the vector store and return the top X matches.

4. Augment the user prompt with those retrieved results, then send the expanded prompt to the LLM for final processing.

Performance and results:

Note: A Mac Mini M4 Pro handled all the computational tasks. No efforts were made to optimise processing. 

Sample set: 11,208 email messages (1.48GB)

Text extraction: 192 messages/sec, 6,488,707 words

Embedding generation + storage:  11 messages/sec

Resulting vector database size: 95.7 MB

Queries: A typical vector query takes less than one second—it’s unbelievably fast. The hit rate is excellent for the kinds of messages you’d hope it would find.

Conclusion:  The results were fantastic. Embedded retrieval is extremely fast, and when it functions effectively, it feels a bit magical.

So what’s next?

We learned a lot building this prototype. However, we understand that HCL has RAG support planned, and since we have no idea what that is going to look like, for now, we’ll wait to see what is included with Domino 2026. Once that is clearer, we will decide whether it is worth investing more time into this concept.

I think that's it for 2026. All the best over the holiday break—and we'll catch up in 2026.