What can happen when you build stuff you find interesting.
My journey building the SkyGlass MCP and what it may help you do.
Today’s post is a radical departure of what I usually share… instead of going for polish and professional prose, I’m doing a mostly unedited stream of consciousness upload of my personal journey. A 100% organic fully human short story of my journey building aviation intelligence tools in the time of AI. So here goes… Clear prop!
SkyGlass happened because I’m a pilot, a programmer and love all things digital + 3D. When the ADS-B mandate came through I saw the potential to build SkyGlass as the first of it’s kind 3D flight tracking platform. That was 2016. Six years and 100k lines of hand-written code later, SkyGlass was born.
Then I found Monkey Werx and started to tune into the world of aviation intelligence. He inspired me a lot TBH. Watching his weekly sit-reps gave me a window into the mil-ops world and tracking fleets in real time with SkyGlass delivered the richness I had hoped for. This inspired the massive SkyGlass watchlist capabilities, the aircraft database integration and more ... but immediately I wanted to venture into 4D tracking … adding the dynamic dimension of time … breaking from now into a new realm where trend and real intel emerged. That was the dawning of the Flight History Reports which are now the core of SkyGlass Pro, and what will soon be available to a wider audience with SkyReports (which I mentioned in my last post).
The turning.
Then AI started to pop off and I began running reports and dumping the exports into my chat bot for analysis… this was an interesting step, as despite the occasional hallucination, a lot of new dimensions started to become visible.
But, being a programmer and loving to automate stuff, I kept thinking there ought to be a way to build something which would free me from the manual churn of running reports and data dumps. I could see the same pain points with my users and online.
The build.
After what seemed like an eternity (time compression in the epoch of AI acceleration makes even a few months of research feel like years, am I right?) I settled on building a derivative dataset and API which is now the SkyGlass MCP. That was in mid June.
After a few weeks the bones were in place and wow… even the early versions blew me away. I still find it amazing actually. Running a 3 hour Flight History report was condensed to seconds, and while I don’t get a killer 3D flight trace visual (which has an intuitive value all its own), I can get higher level trend data in a heart-beat.. which daisy-chains into more and more possibilities.
The now.
And, well... I still gotta pay the rent, so, I made sure what I’m building is suitable for answering real-world, business aviation questions hoping I could help professionals tackle aviation intelligence questions. As an example, see the attached Fall 2026 Business Aviation Report which was built using the SkyGlass MCP and a few round of my AI in tuning the visual output. All in, the data gathering took 8 minutes of tool calls to the MCP.
What would have taken weeks of manual effort and focus was reduced to 8 minutes. I’m still amazed.
The Macro-Scope.
This is not a live tracking widget. Real time tracking is now a commodity, and I see more platforms popping up all the time, both ADS-B feeding networks and real-time dashboards. The SkyGlass MCP answers the trend questions.. it’s a more of a telescope than a microscope… I guess more accurately I would call it a macro-scope (is that a real word? is now!). One of the more interesting derivates available is the Pattern Of Life query, which again reduces an investigation which would have taken me hours to a few seconds. I routinely track all private charter aircraft, and a fair amount of warbirds, but if you task for a PoL on a new bird, the system will automatically enrich and start to maintain the PoL, which only take about a minute.
There are some other endpoints that return h3 data.. data structured around hexagonal grids, allowing for very interesting heat maps like these:
Another example of a report level output is this quick investigation I did when I noticed a burst of HBAL launches in Aug. First I ran a 10 day Flight History report in SkyGlass…
Then I asked my AI to dig a little deeper, and got this…
So, that’s a small taste of what it can do. I’m sure you have your own angles which may be more interesting than what I’ve presented. I have given a select few folks really early access, and the Joint AEROSINT News Network had this to say about the MCP…
“The new SkyGlass MCP is a game-changer. Baseline on an aircraft in one query instead of the usual multi-tab hunt. I’m excited to keep pushing it!”
You want it? You got it.
As a subscriber to my website, you get early access while I'm in the home stretch of releasing the MCP. Stand up a free account and kick the tires. Feel free to reach out with questions or feedback. If you have an investigative angle and you want to share the results online, message me and I’ll give you 1M credits to get it done. All I ask is you’ll tell me how it went and give proper attribution when it’s public.
Clear skies and tail winds,
Goa
Comments