How Artificial Insights is made
Artificial Insights is produced by a pipeline of AI agents I designed, prompted, and tuned over months. I edit the content by hand before each issue ships. This page walks through exactly what the machine does and where I sit in the loop.
Pipeline
- 1
Source
The pipeline pulls fresh articles from 15+ vetted publications — McKinsey, BCG, BIS, the Federal Reserve, MIT Technology Review, Finextra, PYMNTS, Banking Dive, and others.
- 2
Filter
A topic-extraction step scores each article for relevance to AI in banking and discards junk. Typically 80%+ of candidates are dropped here.
- 3
Read & synthesize
Claude reads the surviving articles end-to-end, identifies the top developments, extracts a key number, and drafts the executive summary, per-story analysis, and a strategic insight tying it all together.
- 4
Edit & approve
I read every briefing before it's published or emailed. I rewrite anything that's off, cut anything weak, and kill the whole issue if it isn't good enough. Nothing goes out without my sign-off.
Where I sit in this
I chose the sources, set the editorial angle, wrote the prompts that shape every section, and decide if the briefing meets my standards. Think of me as the editor of a small wire service whose reporters happen to be language models.
What this isn't
I don't draft the prose myself. If you read a briefing and it sounds like me, that's because I shaped the system to sound that way.
Get smarter about AI in banking
Really understand AI in banking: what's real and what's vaporware.
Every Tuesday & Friday • 3 min read • Unsubscribe anytime
By subscribing, you agree to our Privacy Policy