Start here.
This site turns fast-moving AI developments into usable analysis. It explains what changed, what the evidence shows, and what builders or operators can do with it.
What this site is
Rob Allandale tracks primary AI sources: company announcements, research papers, model cards, product docs, changelogs, demos, and real practitioner evidence. The focus is evidence that changes how people choose tools, design workflows, govern agents, or decide what to build next.
There are three types of content:
- Briefings — concise, sourced analysis of specific AI developments. The fastest way to stay current without chasing every launch.
- Essays — longer takes on AI politics, workplace governance, and the practical decisions that matter.
- Lab notes — field notes from running local models, agents, and experimental setups.
Where to start reading
If you want context on what this site watches and why, start with Why this AI analysis site exists.
If you want the political economy of AI at work, read Productivity for whom? The political split forming around AI.
If you want practical guidance on using AI safely in an office context, read The office AI question is mostly a governance question.
For the latest briefings on AI agent platforms, workflow design, and infrastructure shifts, browse Briefings.
How to follow along
The best way to keep up is RSS. No algorithm, no tracking, no forced subscription.
Subscribe via RSS — paste the feed URL into any reader: Feedly, NetNewsWire, Reeder, NewsBlur, or any RSS-capable tool. The feed includes all briefings, essays, and lab notes.
Prefer email? Send a note to [email protected] with "Monthly digest request" in the subject and Rob will add you to occasional roundup emails. No spam, no marketing cadence. Just a monthly digest when there is something worth pulling together.
Process Digitiser
Beyond the public research, Rob applies the same evidence-first method to private business workflows through Process Digitiser.
Upload your process evidence pack, spend 30 minutes with Rob, and get a practical AI-accelerated fix plan in 48 hours. The method applies source-led analysis to one broken workflow at a time.
If you have a workflow that is too slow, too manual, or too dependent on one person knowing the trick — start the intake. Or download the Process Digitisation Checklist to run a quick self-audit first.
About Rob
More background on the methodology, source policy, and correction process is on the About page. Tips, corrections, and interesting AI developments worth covering go through Something I missed?