Day 61 of #100daysofnetworks
Pivot to AI Engineering: Living Systems and GraphRAG from Scratch
Hello everyone! Today is day 61, our first day after completing the baseline Artificial Life Knowledge Graph that we will use for our upcoming efforts! The whole universe is wide open. We can do whatever we want with this Knowledge Graph!
I’ve decided that we should use this opportunity to pivot into AI Engineering!
Networks are used by nature in many ways, and our own intelligence uses networks as infrastructure. So, it is no surprise to me that networks and graphs are important in creating Artificial Intelligence.
I began building Language Models years before there were Large Language Models, and I have done a lot of work with LLMs, so I am excited to bridge the gap between Network Science and Artificial Intelligence, and see what good emerges from this!
Upcoming Training
First, Packt has let me know about some upcoming Machine Learning training! This sounds great, and you can read more on the webpage. This looks like a great opportunity to learn more about the internals of Machine Learning and Artificial Intelligence.
Applied Mathematics of Machine Learning
Saturday, January 24, 2026
Promo-code for a flat 40% off: DAVID40
Tickets priced at $172.56
Blog and Company Updates
There is a lot going on, so let’s start with some high-level updates!
New Books Being Written
New Training Being Created
My company, Verdant Intelligence, is working on developing specialized training. We have identified potential platforms to place the training sessions, and I am currently working on curricula and material. I am writing books and creating training for the following:
Graph: Zero to Hero (Books and Training)
OSINT Wonderland (Training)
AI for Executives (Book and Training)
It is important to focus in order to accomplish goals, so I am beginning with Graph: Zero to Hero. I will write more about this in upcoming articles, but for now it is enough to say that:
The book series plan has been made. This will be a three-book offering (nontechnical, code, code-workbook).
The book outline has been made, because they all share the same outline, in a different way.
The nontechnical book will be an expansive high-level “so what” book. Why are networks and graphs important? Zero to Hero, but in thinking rather than code.
The code book and code workbook will help programmers gradually go from no graph knowledge or abilities to being able to create reliable AI interfaces from scratch.
I will try to open up early access, for readers who don’t mind reading unedited material. I will say more, later.
Graph: Zero to Hero will contain essential knowledge for AI Engineering. I am working on writing Chapter One, today, after this article is written and published.
Blog Pivot: GraphRAG from Scratch
Moving on. Now that our Artificial Life Knowledge Graph baseline build is completed, I am going to pivot this blog to AI Engineering. We are going to work on GraphRAG from Scratch! How fun!
We are going to do this iteratively, from no code whatsoever. The Knowledge Graph is our starting point. We are going to do the work of creating GraphRAG from scratch, because there is no better way to learn than by doing the work. This way, we will be able to focus on the individual steps, and work on them individually until they are good!
Here is a rough plan of the work we will do:
This is the MVP (Minimally Viable Product), and we will apply KISS (“Keep It Simple S____”) and YAGNI (“You aren’t gonna need it”) in our work.
MVP: It has only what it needs to satisfy the goal
KISS: It has only what it needs to satisfy the goal
YAGNI: Any other nice-to-have: “You aren’t gonna need it.”
These three things help with tight focus. MVP KISS YAGNI.
Each one of the rounded rectangles is one task in the approach, suitable for one article or one work session. This allows us to take our time, go slow, learn an approach, and to make adjustments.
The first two rounded rectangles can go together into the first article, so it looks like this will cover at least five articles, but maybe more.
Funny thing: This isn’t new to me. I was figuring out Graph Retrieval before there were LLMs, and I have been creating Language Models since before there were LLMs. I have memories from 2022 that are very similar to what I am now hearing about RAG and GraphRAG in 2025. This is going to be fun. It is familiar territory.
Blog Evolution
Today is Day 61 of this blog series, which is pretty heavy to think about. Each one of these articles has had so much thought behind them, and I prioritized different things at different times. Let’s see what I was talking about on milestone days.
Day 10: Created the Wikipedia Content Crawler (Learn with interesting data!)
Day 20: Explained “Text as Data”, PDF to Text (Crucial for AI Engineering)
Day 30: Explained Word Graphs (Information, Books, Songs, etc.)
Day 40: Created Artificial Life Topic Convergence Graphs (Cool!!!)
Day 50: Found Cognee and a path towards Knowledge Graph Learning
Day 60: Artificial Life Knowledge Graph build complete; ready for AI Engineering.
This entire time, I have been leading you towards INTELLIGENCE. I have shown how to get raw material (crawl, text as data). I have explained how to get to the essence of information itself (bursts, word graphs, dozens of articles), I have shown how this relates to AI Engineering, and now we are going to do AI Engineering, using our Artificial Life Knowledge Graph.
What do you think the future will look like?
Day 70: Reliable and useful Artificial Life chat interface working
Day 80: Living System Intelligence interface designed
Day 90: Living System Intelligence chat interface working
That’s not even a stretch. That’s an easy plan. I will probably hit those goals, early, because my own Living Systems are online and I am actively building out their AI interfaces and capabilities. This isn’t a mystery; I know what to do.
What a year. I might even do a few more articles before 2026 arrives.
Thank you all, for following along! I have really enjoyed getting to know readers who have interacted with me and my articles!
Keeping the Momentum
I wish you all a Merry Christmas and Happy Holidays. I hope you get to do the things that you want to do, and spend time with the people you want to be with.
I am going to keep pushing forward with this blog this week, because it makes me happy. I am with my family, and there’s nowhere I need to be. This is a nice, calm, quiet time for learning and exploration (graph, ideas, etc.).
That means the first of the GraphRAG from Scratch articles will most likely be out in the next few days, and I will probably make even more progress before the end of 2025. I enjoy this work, and I’ll keep at it! Thank you for reading and following along on my adventures!
What’s Next
We are going to focus on AI Engineering for a while. We are going to build a GraphRAG from scratch, only using a Graph Database. This will be a MVP, and we will use KISS and YAGNI to stay focused.
If you would like to learn to create intelligence from scratch, this is the place to be.
Please Support this Work!
I have written over 50 articles for this series. Each one takes about four hours of research, and several pages of writing and editing. Here are some ways you can support the blog!
Please subscribe if you have not. This motivates me like nothing else!
LET’S DO BUSINESS. Reach out to me if you need data or AI help! Happy to help! You can read about the partners we are currently looking for here.
BIGGEST HELP to the BLOG: Please consider upgrading if you are a subscriber. Thank you to all current paying subscribers for making this research and development possible!
Please buy my book to understand how I think about Natural Language Processing and Network Science combined.
Please reach out to me if interested in training related to Data Science, Data Engineering, Network Science, Knowledge Graphs, Artificial Intelligence, or anything else I write about. Feel free to message me on Substack or on LinkedIn!
Feel free to hang out in the comments and have a good time!
We have come so far since the very first day of the very first #100daysofnetworks. I love writing for this series. Thank you for being a part of it!
Love the MVP approach for something as complex as GraphRAG. The temptation with these projects is always to build everything at once, but breaking it into the five rounded rectangles makes it way more managable and teachable. I built a similar retrieval system for a healthcare startup last year, though we ended up using vector embeddings alongside graph structure. The holeup was always entity resolution when new content came in, we kept getting duplicates until we addedcontext-aware merging. Excited to see how the graph-only approach handles that problem without leaning on embeddings as a crutch.