Halifax, Canada / Crashboard
Andrew Davies
A public notebook on AI work, research systems, and strategic judgment.
I publish the parts of my working notes that are useful outside the notebook: AI workflows, source-backed research methods, knowledge-system design, and strategy writing from Halifax.
Crashboard is where finished-enough thinking leaves the private notebook. The wiki holds the current corpus: source-backed pages, recurring concepts, and the connective tissue between them.
The blog is for questions that need a fuller answer than a wiki page can carry: how a workflow works, what a source trail proves, and where the judgment is still uncertain.
Read the fuller profileLatest briefs
Current signals with the source trail intact.
Each morning brief identifies consequential developments, explains why they matter, and links back to the original reporting and deeper wiki context.
Research system
From a daily signal to durable context.
The public notebook is organized in layers so readers can scan what changed, follow a recurring question, or inspect the deeper synthesis behind it.
- 01
Morning briefs
Dated, source-backed scans of developments worth tracking across AI, defence, infrastructure, markets, and institutional change.
- 02
Topic hubs
Durable research trails that connect recurring developments and make the archive useful beyond the day each brief was published.
- 03
Public wiki
Deeper synthesis, concepts, operating models, and source notes connected through a searchable knowledge graph.
Wiki
The real public content lives in the wiki.
These pages are generated from the compiled knowledge base and link to real public routes.
AI Automation Builders
An AI automation builder is a workflow-first operator who connects LLMs to real business tools, rebuilds repetitive processes as reliable pipelines, and sells measurable business outcomes rather than frontier-model novelty.
AI Foundations & Model Adaptation
AI systems become valuable when broad model capability is turned into useful behavior through architecture, adaptation, grounding, routing, and surrounding workflow design.
AI Safety & Control
Safety is not one feature bolted onto a model. It is a layered control problem spanning training data, model behavior, prompt design, runtime checks, retrieval policy, user permissions, organizational governance, privacy risk management, evaluation quality, infrastructure resilience, orbital and terrestrial service continuity, and the human capacity required to supervise and collaborate with those systems well.
AI, Agents & Software Systems
This is the domain map for AI, agents, coding workflows, software systems, model foundations, memory, safety, and verification.
AI-Assisted Content Systems
AI-assisted content systems are personal publishing engines where capture, note-linking, retrieval, prompting, and performance feedback compound over time so writing starts from a rich vault instead of a blank page.
AI-Native Organizations
AI-native organizations are not defined only by using AI tools. They are defined by redesigning work, incentives, interfaces, team structure, and human capability around the fact that intelligence and execution can now be delegated much more cheaply.
Next
Start with the material that already has weight.
The wiki is live now. The blog carries the longer essays and daily briefs as the archive grows.