This release brings one of the most “useful “lab” updates I’ve shipped so far:
a full-fledged logging and monitoring system for LLM API calls.

Inside the new Logging tab in the settings, you’ll now find:

• Detailed usage charts showing API call volumes, response times, and errors.
• Interactive filters for provider, model, call type, and date range.
• CSV export for those who like digging into the numbers.
• Configurable log retention (default: 60 days) with automatic cleanup.

So why this feature?

Because I wanted a clear way to track and verify LLM usage, monitor performance, and get a feel for latency differences across providers. It’s also a foundational step toward something bigger — turning AIDA into a local hub for LLM calls and, eventually, a lightweight agent layer (think CrewAI, but simpler and fully client-side).

Connecting the dots: AIDA + my other tools

My dream setup is to have AIDA quietly coordinate across my other extensions — a sort of glue layer for local AI workflows:

• InBuddy, which helps with generating, summarizing, or understanding LinkedIn posts (the first real-world use case that got this ball rolling).
• Tab Deck, for automatically classifying a browser deck or suggesting better tab ordering when your workspace becomes digital chaos.
• ORGanizer, the Salesforce companion tool that helps navigate, document, and understand complex orgs.

And that last one — ORGanizer — deserves a closer look.

ORGanizer, security, and the “use at your own risk” principle

Salesforce data is sensitive, and ORGanizer operates in that high-trust environment. So when I talk about AIDA integrating with it, that doesn’t mean a free-for-all of AI inside your CRM. Quite the opposite.

The goal is to reduce the permissions footprint, not expand it.
AIDA could act as a local interpretation and summarization layer — analyzing metadata, providing insights, or suggesting optimizations — but without data ever leaving your environment unless explicitly allowed.

Think of it as an optional intelligence layer that can assist you in understanding your org’s structure or dependencies, while fully respecting the strict security boundaries companies enforce.

This is the kind of feature I’ll explicitly tag as “use at your own risk”, and one that I expect many security teams will choose to centrally enable or disable depending on their policies. AIDA’s design already anticipates that kind of governance — flexible, transparent, and privacy-aware.

The ongoing experiment

AIDA remains a living, breathing experiment — coded in the moment, with features appearing as they cross my mind. There’s no strict roadmap, just a curiosity-driven evolution.

I’m always open to suggestions and feedback. You can follow the whole AIDA story and all its strange evolutions on the AIDA feed on my blog.