Software engineer shipping a real product with coding agents. I share what works, what breaks, and which models are worth it. Building @luciditybot.
Madrid
Joined December 2009
- Tweets3.3K
- Following1.4K
- Followers5.7K
- Likes18.5K
Pinned Tweet
I completely stopped using Codex app.
Pi + Herdr + Collie + Executor is unbeaten.
Is not even close.
The only thing I need to figure out is Scheduling tasks. Any advice ?
Jose Mejias retweeted
Spent multiple hours now evaluating Railway alternatives to move my shit to. Conclusion is that Railway's product is uniquely good and nothing comes close to a drop in replacement 🙃
Minimum cost to replicate my setup today is 3x higher than I am paying them currently
what's going on with Railway this morning? this is so annoying
yesterday builds were also failing :(
Jose Mejias retweeted
¡Hola, Region of Madrid! 🇪🇸 We’re thrilled to announce that we’re coming to Spain, beginning with Madrid. This will be our first region in Spain and second in the European Union. We plan to launch fully autonomous ride-hailing in Madrid in 2027.
Sign up for future updates: waymo.com/waymo-in-spain
¡Hola, Comunidad de Madrid! 🇪🇸 Nos emociona anunciar que llegamos a España, empezando por Madrid. Esta será nuestra primera región en España y la segunda en la Unión Europea. Planeamos lanzar nuestro servicio de ride-hailing totalmente autónomo en la Comunidad en 2027.
Yesterday i setup Omarchy on my old gaming PC from 2020 then added it as a T3 Code environment, and its only 9% slower in my CI workflow compared to my Macbook M2 Max. Feels good unlocking more compute for "free".
Surprisingly fast!
Jose Mejias retweeted
🚀 DeepSeek-V4.1-Flash-1017 is here!
Stronger benchmark performance, faster task completion, and shorter outputs than the previous version—a Pareto improvement in performance and cost. Now live on our website, app, and API.
🛠️ Optimized for @DeepSeekHarness, Tencent CodeBuddy, WorkBuddy, and more. Upgrade to DeepSeek Harness v0.2.1-rc.1 or later for the best experience with the new model.
1/4
Jose Mejias retweeted
what's this?
Jev is evolving!
tune in to our town hall with @CompleteSkeptic and @allietheicon at 1:14 PM to learn more
If you are like tinkering, tell your agent to setup WhatsApp MCP on a computer you can leave always on.
Run a Tailscale funnel on the MCP server.
Sign up to @executorsh and tell your agent to add the MCP as an app.
Add executor to your agent. Be it Grok Bot, a Dot any anything. Enjoy.
HOW TO CONNECT YOUR GROK @bot TO WHATSAPP
send this tweet to your Grok Bot so it can handle the setup for you:
1. have it open web.whatsapp. com on its computer.
2. have it choose “link with phone number” and enter the number it already has for you.
3. have it generate the linking code and show it to you.
4. on your phone, open WhatsApp → Settings → Linked Devices → Link a Device → Link with phone number instead. enter the code to approve the connection.
5. have it check that your chats load and confirm it's connected. your Grok Bot can now read messages and draft or send replies on WhatsApp.
once connected, you can have it:
> organize your chats
> flag messages you haven't replied to
> add plans to your connected calendar
> and remind you about things you said you'd do
so between WhatsApp, my email, and X DMs, my Grok Bot now has an absurd amount of actionable context on my life :)
forget about ps5 or photo frames, who's jailbreaking amazon echo dots? i want to use my Alexa speakers with Grok bot, without amazon cloud
Jose Mejias retweeted
Introducing ChatGPT for Dishwashing
GPT-6 achieves state-of-the-art results on dishwashing benchmarks for long-running jobs and difficult stains.
The Navier–Stokes Millennium Prize Problem is a $1 million challenge about whether the equations describing fluid motion can break down. In this video, I accidentally found a counter-example while trying to optimize my dishwasher. But in reality, OpenAI used 10,000 AI agents to find a proof in 88 hours: openai.com/index/navier-stok…
Parody of the original GPT-6 Astra ad
@OpenAI #ChatGPT_Partner #chatgpt
Credits:
Written and Directed by Joma
Co-Writer - Henry Connor Coan IV @coan_henry
Production by Dauntlus Studios @dauntlus.studios
Producer/Co-Director - Marcus Liew @dauntlus
Camera Operator - Amos Lee @amoselijahlee
Art Director - Alethea Soo @aletheasoo
Art Assistant - Alicia Lim @aliciatkl_
Assistant Editor - GPT-6 Astra
Music: Tensions Run High by Soundridemusic
youtube.com/watch?v=Ly9H63SL…
he creado una skill y cli para comprar tickets en @Renfe
pídele a tu agente que la instale:
github.com/mejiasd3v/renfe-c…
computer use no es barato, al menos con los API costs
rollo, simplemente reservar un billete de tren en la página de Renfe usando Sonnet 5.5 o Sol 6.1 Sol veo difícil bajarlo de los 50 céntimos (!!) y aún así fallan
estaría la gente dispuesta a pagar 50 céntimos por no usar la página web de la Renfe para reservar un billete? yo sí porque la odio con todas mis fuerzas, pero no se yo el resto
las compañías están literalmente quemando dinero con los asistentes con el objetivo de conseguir cuota de mercado
supongo que la apuesta es que los modelos van a mejorar rápidamente de tal manera que el OPEX caiga muy rápido
pero a día de hoy yo creo que están QUEMANDO billetes
Jose Mejias retweeted
I also have a little announcement for DevDay. I asked Tibo to include it in the presentation but he didn't read my slack DM 😭
My first onboarding project is merged. I fixed this issue once and forever! I reduced amount of disk writes by 50% and the size of stored files by 4x
Jose Mejias retweeted
I finally tested GPT-6 Sol on a real work. 2 repos. 105 hidden bugs. Find and fixed what you can.
It looks like a huge degradation. The results:
- GPT-6 Astra (max): 45
- GPT-5.6 Sol (max): 43.5
- Opus 5.5 (max): 41.7
- Muse Spark 1.3 (max): 32.2
- GPT-6 Sol (max): 29.3
Until you measure the cost (API-equivalent):
- GPT-6 Astra (max): $33.04
- GPT-5.6 Sol (max): $95.25
- Opus 5.5 (max): $58.53
- Muse Spark 1.3 (max): $18.11
- GPT-6 Sol (max): $9.93
More effort levels (xhigh, high, medium, low) dropping in this thread today 🧵
