LaegnaAIBasics is an introduction to deep learning AIs that stays as close as possible
to “classical craft” understanding. It begins with the perceptron – a single decision
unit – and shows how stacking these units into layers creates a network that can
recognize patterns across images, language, and other data. Instead of drowning
you in formulas, it walks through tensors, matrices, and gradients with analogies
to weight, leverage, projection, and reflection, so the ideas can be shared by anyone
willing to think carefully, regardless of background or gender.
From there, it introduces invariants – the features that stay meaningful when inputs
change – and discusses how to visualize high‑dimensional spaces, so you can “see”
what your AI is doing instead of trusting it blindly. The goal is that once you
understand this one “wheel” of a perceptron and gradient descent, you can recognize
the same principles in convolutional nets, transformers, and beyond: bicycles,
cars, and carriages of the AI world built from the same circular core.
Includes: AI intro, invariants, visualization, compassion
LaegnaPracticalAI shows how to turn your ideas, notes, presentations, web pages,
and media into a structured document collection – a living workshop where AI can
actually work with you. Instead of treating models as magic, it treats your documents
as the main raw material: they define what the AI can know about you, your team, or
your project. The repository is organized into themed folders (ChatProvider,
ChatWithDocs, CommonChat, DocsOwnPresentations, DocsOwnVideoPresentations,
DocsOwnWebpages, Images, and more), each with its own README that explains how to
wire that kind of content into conversations and tools you can actually use every day.
In practice, this means building consistent pipelines for indexing, retrieving,
and presenting your documents, so you can create AI assistants, resource builders,
and knowledge explorers that reflect your own domain. Together with the
docu‑deploy‑ai website, it embodies the idea that everyone can own an AI
workspace – not just people with big infrastructures or formal data warehouses.
Includes: chat with docs, multi‑source collections, deployment patterns
LaegnaAITraining is about what happens after you have a functioning document
collection and a basic understanding of deep learning: you begin to train models
so they do not just reference your documents, but grow an internal sense of your
style, priorities, and theory. In the same way a craftsperson develops an instinct
for wood grain or stone, a trained model can develop an internal “grain” of your
work – one that helps it generate, summarize, and reason in ways that feel less
generic and more like a genuine collaborator.
The repository focuses on how to prepare your own material, how to structure
training so it remains interpretable and ethical, and how to think of AI as a
contemplative partner: something that revisits your texts, refines its understanding,
and becomes better attuned to you over time. This is where researchers, writers,
and builders of any gender or discipline can move from “using an AI” to “cultivating
a shared language” with it.
Supports: deeper alignment with your voice & ideas