dbuxton

All the travel apps out there are built for the instagram generation but dinosaurs like me still use TripIt for travel tracking and itinerary sharing. But its parsing hasn’t changed since Concur bought it 15 years (!) ago and I thought it would be nice to have LLMs read the cancellation minutiae for trips and put in a simple timeline for sharing.

Comment · 2 months ago

Many runways have something called PAPI (precision approach path indicator) which does something similar. Basically you have four (or two sometimes for less precise version) lights which appear either red or white depending on whether you are high or low on the glidepath and you aim to get two reds and two whites - more whites means too high, more reds too low.

I can see that this would be not very intuitive for left/right guidance but are similar systems used in ship navigation?

Comment · 2 months ago

Well I just rent so no idea. Not sure they will end up lower as presumably it will all need to be conducted by a FAA double-starred technician who is not just vibe coding it

Comment · 2 months ago

27 mins endurance (?) for $5 of power. Not bad. I rent a 172 and have been paying a $5/h fuel surcharge for the past few months due war in Iran - but at UK avgas prices 27 mins in my plane would be $55ish. Haven’t looked into the details but kind of crazy impact on economics of aviation if this size plane (10x 172 MTOW) can actually do useful short haul

Comment · 2 months ago

There's no such thing as a "timeless" translation - if you read something written by Victorians you'll get all sorts of overlaid assumptions and biases, just ones that are unfamiliar

Comment · 3 months ago

https://chatgpt.com/share/6a673180-0a78-83ed-abd0-d797c4d95a...

tldr the initial effort didn't seem to read the linked article so I copied and pasted the full text in

Comment · 3 months ago

I can’t read something like this without thinking of the late great Richard Scarry

Luckily ChatGPt does a pretty good impression https://chatgpt.com/s/m_6a6662e9af6c81918523fb06d1384e74

Comment · 3 months ago

At the moment a lot of companies thinking of it like speculative capex - we can just stop doing it if we don’t get results, or we can easily optimize.

My concern is that with all the bundling that the model labs are doing the lock-in becomes harder than anticipated to unwind

Comment · 3 months ago

It is just extremely difficult in most companies to draw a line between a specific feature and an amount of revenue. A lot of engineering has a stochastic type of impact.

(It’s often easier to see what a feature does to reduce costs though)

Comment · 3 months ago

As well as cost-per-task I think it's worth thinking about speed, especially in non-coding contexts that benchmark less cleanly

We've started trying to do some comparison videos to capture more of the UX vs speed vs cost stuff e.g. https://www.linkedin.com/feed/update/urn:li:activity:7479891... which one of my team did for my LinkedIn account (disclaimer: marketing)

(In this particular case Deepseek was way slower than GPT 5.5 but I think that's because it installed Libreoffice half-way through the task!)