Recently, I've been so eager to get new model releases in Codex. I'm hooked. I hope this accelerates development. Shows how dependant I have become to Codex.
I run a SaaS that allows for court bookings. A tennis club has exactly this scenario: club offers 1h, 1.5h and 2h bookings, but 30m start. Since it is a private club and requires a membership, gaming the system is not a real issue. However, they can never be fully booked because of the 30m holes.
They open at 7am so in the mornings bookings are on the hour. Competitive junior athletes start at 2:30pm shifting to half-hour starts in the afternoon. So they mess up their schedule to accomodate them.
The non-optimal solution is to avoid proposing slots that would create 30m holes. For example do not offer 5pm if a booking ends at 4:30pm. We implemented that but it does not solve the fundamental issue. And they don't want to change. Such is life!
What is not demoed are basic AI capabilities I need today to effeciently use my iPhone. Examples:
- Siri, send the latest 20VC podcast to my watch. I'm going jogging.
- Siri, do not start Apple Music. I do not use that app. Start Podcast.
- Spell check and Auto-Correction on iMessage should support French and English. I switch from the 2 all the time.
- Etc...
I came in thinking they would be like PE and just put products on life support sucking all the recurring they can. But it seems they care and improve the products. I think that has merrit.
In my last jobs Jira was used, and despised by all except product managers. It just becomes a mess.
In my startup (now 20 people), we use Trello. Outsiders look at us funny. I respond that its the same company after all...
In Codex I was suggested to try Codex Spark for a limited time. So for my next session, I gave it a shot.
It is much, much faster. However on the task I gave it, it spun around in circles cycling through files and finally abandoned saying it ran out of tokens.
Major fail.
Currently integrating an AI Assistant with read tools (Retrieval-Augmented Generation or RAG as they say). Many policies we are writing are providing context (what are entities and how they relate). Projecting to when we add write tools, context is everything.
They exist to protect the creator/inventor and allows them to get an ROI on their invested time/effort. But honestly today, the abundance of content, especially that can be generated by LLM, completely breaks this. We're overwhelmed with choice. Content has been comodotized. People will need to come to grasp with that and find other ways to get an ROI.
The article does provide a hint: "Operate". One needs to get paid for what LLMs cannot do. A good example is Laravel. They built services like Forge, Cloud, Nightwatch around open source.
Most analysts seem to forget what actual consumers do. Normal people use ChatGPT. They accidentally use Gemini when they Google something. But I don’t know anyone non-technical who has ditched ChatGPT as their default LLM. For 99% of questions these days, it’s plenty good enough—there’s just no real reason to switch.
OpenAI's strategy is to eventually overtake search. I'd be curious for a chart of their progress over time. Without Google trying to distort the picture with Gemini benchmark results and usage stats which are tainted by sheer numbers from traditional search and their apps.
A company replacing domestic workers by cheaper H1-B workers. As opposed to a company shutting down because foreign competitors took their marketshare.
In either case, domestic company is not competitive. Protectionism won't make the domestic company more competitive.
UI, in my mind, has always been the hardest part of the stack. It is even more true today. A front-end dev must:
- Deal with a subjective human (not an objective machine)
- Deal with designers and product managers
- Be constrained by the back-end engineer and infrastructure
- Make it work on multiple devices and browsers
- Use the framework someone else chose
- Not break anything