Start with a real AI problem. Follow it through the mechanism, evidence, capstone implementation and inspectable code to a solution you can actually use.
Programmer.ie solutions are the practical destination of the research, books and experiments.
A book should not end at explanation. Each finished book is expected to lead to a solution: a durable page organised around a real problem, a concrete outcome and an implementation that can be inspected. The book earns the mechanism, the capstone integrates it, and the solution becomes the applied home for the work.
flowchart LR
R[Research] --> B[Book]
B --> C[Capstone]
C --> S[Solution]
S --> G[Repository]
G --> U[Real use]
U --> E[New evidence]
E --> R
A solution can also begin somewhere else. A strong blog post, experiment or engineering investigation can grow into a solution when it becomes comprehensive enough to answer the problem rather than merely discuss it. The solution page then becomes the canonical home, while supporting posts provide narrower entry points into it.
Start with the problem Browse the solutions below by domain or featured project.
Understand the mechanism Each solution connects back to the book, experiment, or underlying technical argument.
Inspect the implementation A real solution leads to code, experiments, tools, or a live system.
A solution on Programmer.ie is not just an article and not just a repository.
It is the place where:
Every mature solution should make five things easy to find:
Some solutions below are live systems. Some are being built directly from current books. Others will appear as the library grows. The goal is not to manufacture a product for every chapter. It is to make sure important technical work eventually terminates in something concrete that can be built, tested, inspected or used.
OpenCode Remembering is the working application behind Memory From First Principles: a standalone plugin and bundled memory engine that turns project history into controlled, provenance-bearing context without treating everything retrieved from the past as permission to act.
View solution →ZeroModel is a working system for building AI decision flows that preserve enough structure to inspect what was claimed, what evidence supported it, what policy allowed it and what actually happened.
View solution →RELATE is the capstone system for Embeddings From First Principles: a small runtime, corpus, benchmark suite, and evidence ledger for asking what an embedding space preserves before a system is allowed to use it.
View solution →CodeAI is the capstone system for Applied AI: an experimental runtime that treats models as replaceable cognitive components while the surrounding software owns state, evidence, authority, side effects, verification, completion, and what happens next.
View solution →AIbussin turns the Freestyle Cognition research into practical workflows, prompt builders and interactive tools that help people explore, reframe and develop ideas with AI.
View solution →The DSPy From First Principles capstone: a bounded repository-repair system where the model proposes diagnoses and patches, while software controls tools, validation, holdout evidence, comparison and promotion.
View solution →Pi Patch Desk is the runnable capstone for Pi Agents: Building on the Coding Agent. Scripted responses drive the actual Pi loop against a disposable TypeScript fixture, while the host checks the patch and records the evidence for your review.
View solution →Digital Lens is the Browser AI capstone direction: a browser-native assistant that helps the user control attention and information flow instead of surrendering those decisions to each website or AI provider.
View solution →A reusable Python pipeline for turning arXiv search results into a local SQLite corpus of PDFs, extracted text, page chunks, and embeddings.
View solution →A compact Python pattern for combining SQLite document storage with a FAISS nearest-neighbor index for local RAG retrieval.
View solution →A small local RAG retriever that stores text and vectors in SQLite, searches with both FTS5 and sqlite-vec, then merges rankings with RRF.
View solution →A reusable PostgreSQL pattern for AI applications that need embeddings, relational metadata, indexes, and semantic nearest-neighbor queries in one database.
View solution →A minimal local tool-calling loop for Ollama: register Python functions, pass schemas to the model, execute selected calls, and feed tool results back into chat.
View solution →A modular document intelligence pattern that turns papers into structured, domain-tagged, scored knowledge for retrieval and self-improving AI workflows.
View solution →A reusable agent pattern for replacing single-path generation with scored reasoning trees that can be inspected, compared, and improved.
View solution →A reusable media pipeline for converting long-form research audio into structured transcript chunks, visual prompts, generated images, and a synchronized video.
View solution →A reusable case-based reasoning pattern for agents: treat past PlanTraces as searchable cases, rank them by similarity and quality, adapt the best ones, and record reuse lineage.
View solution →A model-agnostic evaluation pipeline for turning score gaps between two models into calibration curves, routing thresholds, visual diagnostics, and provenance.
View solution →A verified self-play loop for agents: proposer creates tasks, solver answers with evidence, verifier checks support, and metrics become reusable training signal.
View solution →A small Python validation primitive for checking headers, required fields, types, allowed values, and maximum lengths in tabular files.
View solution →A visual AI pipeline for comparing populations of reasoning traces: gather good and contrasting examples, render VPM timelines, subtract fields, and rank the metrics that separate them.
View solution →A modern Python replacement for a 2016 VBScript workflow that rasterized SVG artwork into Android drawable and mipmap density folders.
View solution →A dependency-light Python pattern for writing a static lookup file once and reading it many times through an indexed key map.
View solution →