Benjamin Houdu (@benjaminhoudu) header
Benjamin Houdu (@benjaminhoudu) profile picture

Tech Lead at https://xitter.cf/t.co/kNEKmSsidI Opinions are my own and not the views of my employer

Paris, France
Joined June 2007
Photo by Benjamin Houdu (@benjaminhoudu)Photo by Benjamin Houdu (@benjaminhoudu)Photo by Benjamin Houdu (@benjaminhoudu)Photo by Benjamin Houdu (@benjaminhoudu)
Merci à tous ceux qui sont venus voir ma présentation du serverless chez AZ Trade à l'AWS Summit Paris :)
2
238
Benjamin Houdu retweeted
Jensen Huang est carrément désagréable avec Arthur Mensch. Mais il faut reconnaître que l'anglais de Mensch est *catastrophique*. Enfin merde il a fait l'X puis un doctorat mais il ne maîtrise pas les pluriels ni la prononciation basique. Dans pas mal de pays européens, n'importe quel caissier, électricien ou agent municipal parle couramment anglais alors qu'il ne présente pas à des conférences internationales. Tu es milliardaire mec, prends un prof particulier. 2h par semaine et dans 6 mois ton accent est 10x mieux.
Jensen a été proprement HORRIBLE avec Arthur Mensch lors de cette conférence : je mets la vidéo pour ceux qui doutent, avec les sous-titres 🔽 Ce n'est absolument pas une façon de parler en public. Le mec dégringole dans mon estime.
133
47
11
849
225,111
8
56
3
723
11,825
Benjamin Houdu retweeted
203
8,322
668
86,565
4,578,544
Security teams were told to make lemonade from whatever the year handed them. Nobody mentioned the chute.
3
57
4
424
9,842
Benjamin Houdu retweeted
«Todo el código es basura, excepto el mío». Lauren Tan, ingeniera de GrokBot en SpaceXSI. En el AI Engineer Code Summit acaba de mostrar toda su configuración de desarrollo. Cada error de un agente se convierte en una regla de linting, no en un prompt. Los agentes verifican su propio trabajo con vídeos, capturas de pantalla y tests. ¿El resultado? 3.000 PRs llevadas a producción en un mes. Su base de código está ahora tan limpia que los agentes fusionan la mayoría de sus propias PRs. Mira esto antes de automatizar cualquier cosa.
49
592
39
6,019
518,642
Benjamin Houdu retweeted
Since Apple Music and Spotify designs are so boring, I built an iPhone Duo app called Duoamp that is a throwback to Winamp Shake it and a new Winamp-style skin appears
256
505
134
4,117
248,403
Hehe
6
204
3
3,383
19,973
Benjamin Houdu retweeted
Logo design for Polar Bar
202
1,789
107
75,735
1,077,161
😭🤣🤣🤣
317
4,278
260
49,151
960,466
Benjamin Houdu retweeted
On Microsoft Outage
1,027
4,648
743
51,325
4,436,904
Benjamin Houdu retweeted
As expected, the thing that is blowing people's minds most in my course is not the agentic stuff. It's the DDD stuff - domain-driven design. Building a ubiquitous language, using it with the agent, and refactoring the code to use it. Just bonkers how good it is
217
126
34
3,284
266,686
Benjamin Houdu retweeted
this guy wins again
JUST IN: Vibecoder discovers previously unmapped ancient Maya settlement hidden beneath the Mexican jungle using 48 AI agents & NASA lidar data, despite having no archaeology experience.
Readers added context they thought people might want to know
‪Update (11 November): it turns out the site had already been mapped in Estrada-Belli et al. 2025. The agents missed it while going through the literature, so they rediscovered it rather than discovered it (it just wasn't in INAH's register yet). Still work to do then!‬ ‪pietromiotti.com/latest/maya-si…‬
57
1,051
29
16,938
269,227
Benjamin Houdu retweeted
JENSEN! THATS NOT WHAT WE MEANT BY CHINESE MODELS!!!!!!!
591
8,117
701
122,603
2,699,494
Benjamin Houdu retweeted
Caste system is looking to be true
296
341
74
5,130
588,048
My experience is that it's usually a waste of time to have an LLM go through the same process a human would, no matter how good it is for humans to use that process. I've read a couple of reports of someone instructing an LLM to do TDD, for example. A lot of TDD's value lies in getting us humans to think about the problem (and solution). Since LLMs don't think, I haven't found much benefit in having them go through a process designed to encourage thinking. That said, the notion of writing an initial "test" that provides an example of how we want the code to work, then telling an LLM to write the code to make it pass, isn't a bad approach. I'm in the loop, though. The main problem with that approach is the same one that TDD has—it's easy to lose the forest in the trees and not pay enough attention to overall architecture. I find that ATDD (Acceptance-test driven development) doesn't have that problem, but it's pretty coarse-grained.
21
11
2
99
9,523
Benjamin Houdu retweeted
You never forget the sound of a Windows PC starting up in the 90's
529
2,919
490
20,226
2,927,365
Benjamin Houdu retweeted
Que s'est-il passé ? Faute de frappe, gage perdu ou tout simplement coup de génie. Ils sont 3 joueurs à avoir trouvé le 9 à 1. OUI 3... Nous voulions leur faire une petite dédicace
6
11
1
151
17,284
Everyone's first PHP project started with a friendly mascot and a mysql_connect call.
2
21
1
229
8,873
Benjamin Houdu retweeted
very weird thing happening rn technical people are sad, they think their skills are now useless non-technical people are desperately wishing they were more technical so they could use agents better
1,553
2,410
683
36,887
1,981,084
Benjamin Houdu retweeted
Everyone should focus on automating code reviews. Right now, agents can write a virtually infinite amount of code, but it's for nothing if it doesn't work. I see people clinging to "manually reviewing the code" as the only way forward. I think this is a mistake and a major bottleneck we all need to get rid of. You want people to focus their time on things only people can do. You want agents to focus on everything else. There's no one-size-fits-all solution here, but you should spend as much time as possible building a process that removes the review bottleneck.
164
22
12
266
73,022