Curated developer articles, tutorials, and guides โ auto-updated hourly


Fifty posts. Ten chapters. One framework. The Learn Arc closes with a reader's map, a short what-to-...


Chapter 7 is the muscle chapter. Full-colour POMDPs, belief-propagated tree search, Dirichlet-learne...


The hero claim of the whole book: one variational rule explains both what you believe and what you d...


Chapter 4, ยง3. Chapter 3's abstract G(ฯ) meets the A and C matrices and becomes an actual computatio...


Chapter 4, ยง1. Before you fill any matrices, you list three things: hidden states, observations, act...


The second session of Chapter 1. Open the Tiny Open Goal maze, press Step three times, and watch Per...


A 50-post series teaching Active Inference by building it. Part 1: the map. Native Jido on Elixir/OT...


Chapter 10 is the book's closer. Where does the unified theory go, and where does it bend? Perceptio...


Chapter 3, ยง3. Softmax over โG/ฯ โ and why that ฯ is a meaningful biological quantity, the hinge to ...


The third session of Chapter 1 closes the overview. Why Active Inference deserves "unified theory" b...


Chapter 6's closer. Boot the custom agent, watch it run in Studio, read every signal in Glass. The w...


Chapter 2's second session. Bayes' intractable evidence meets KL divergence. Variational free energy...


Chapter 3's opening. Expected Free Energy in its cleanest form. Risk plus ambiguity. The value of a ...


Chapter 3, ยง2. The two columns of Expected Free Energy, unpacked one at a time. When each dominates,...


Chapter 4 makes the theory concrete. A/B/C/D matrices, Eq. 4.13 belief updates, Eq. 4.14 policy post...


Chapter 6 is the practical how-to. States, observations, actions, A/B/C/D, run, inspect. Every cookb...


Chapter 9 flips Active Inference from theory-of-brains to tool-for-studying-brains. Free energy as m...


Chapter 3: Expected Free Energy. Why one softmax over risk + ambiguity makes an Active Inference age...


The depth arc begins. One 8-minute session from Chapter 1: what Active Inference claims, in four lea...


Chapter 5 lands the theory on the brain. Predictive coding as message passing; ACh, NA, DA, 5-HT as ...


Chapter 2's final session. Let the observation be a free variable and free energy turns from a poste...


The final Chapter 8 session. A continuous-time playground. Change precisions on the fly, poke the wo...


Chapter 2, ยง3. Free energy decomposes into complexity plus accuracy. What "cost" means when your gen...


Chapter 6, ยง2. Fill A (sensor), B (transitions), C (preferences), D (initial prior). Four matrices, ...