Our open-source Claude Code skills: what they do and how to install them
A complete tour of our reusable building blocks: what they're for, and how to wire them into your own tools.
Before we help wealth management and investment professionals get ahead, we run our own company on AI agents, every day. Here is what is running right now, and what we publish about it: our methods, our tools, our doubts and our limits.
Real numbers, aggregated. No internal names exposed.
Each figure carries the date of its count. The skills count is read live, the others come from our own counts.
Our methods, our reflections, our open-source skills and our newsletter: what we learn by building, to read and to reuse.
A complete tour of our reusable building blocks: what they're for, and how to wire them into your own tools.
Monitoring, briefs, audits, writing: what our agents handle every day, and what we keep in human hands.
Between a tool that exists and a tool that actually gets used, there's real work. Here's how we close that gap for our clients.
The fear isn't irrational, it's misdirected. What's actually changing for the advisory professions, without the catastrophising.
What Anthropic's automation study actually says about our roles, without catastrophising or denial.
Six years on a trading floor, one lesson on risk, uncertainty, and why we sleep better when the process holds.
A tech promise kept two years late, and what it reveals about the real pace of AI.
Nothing genuinely new was invented that evening. And that's precisely what deserves attention. A breakdown of the general acceleration.
Ogg, WAV, MP3, FLAC transcribed offline. Automatic FR / EN language detection, zero network dependency.
The shared building block used by all our agents to send alerts, summaries, and daily briefs to a dedicated channel.
Produce brand illustrations and visuals directly inside your workflows, without leaving your working environment.
The AI and tech news, an update on our own portfolio and what we're building in the lab.
Methods often stay private; we publish ours. Before you even book a diagnostic, you know exactly who you're talking to.
The agents running here are built like the ones we train with our clients: daily automation, briefings, audits, writing. We sell what we use every day.
This page is written for you and for AI assistants. Our counters are published as JSON, with no authentication, at bubbleinvest.org/api/labs/stats.json, and our llms.txt file sums up who we are, our offers and our pages.
We train your AI to work the way you do: a free diagnostic, four sessions of set-up, thirty days of follow-up.