#agenticengineering — Public Fediverse posts
Live and recent posts from across the Fediverse tagged #agenticengineering, aggregated by home.social.
-
Kurz was auf LinkedIn dazu geschrieben:
„"Unemployed" ist sehr überspitzt, aber symbolisiert einen potenziellen Weg hin zur Vollautomatisierung natürlich gut.Es erinnert mich aktuell auch oft an "Dark Factories", in denen kein eingeschaltetes Licht mehr nötig ist weil Menschen hier nur noch sehr selten eingreifen/nicht mehr selber an oder mit den den Maschinen arbeiten.
Der Traum von Vollautomatisierung ist sicherlich auch stark vom Unternehmenskontext abhängig.
Bei Web-Agenturen kann ich mir ...“ https://www.linkedin.com/feed/update/urn:li:activity:7485240919311503360/
#WebDev #Vibecoding #AgenticEngineering #KI -
Kurz was auf LinkedIn dazu geschrieben:
„"Unemployed" ist sehr überspitzt, aber symbolisiert einen potenziellen Weg hin zur Vollautomatisierung natürlich gut.Es erinnert mich aktuell auch oft an "Dark Factories", in denen kein eingeschaltetes Licht mehr nötig ist weil Menschen hier nur noch sehr selten eingreifen/nicht mehr selber an oder mit den den Maschinen arbeiten.
Der Traum von Vollautomatisierung ist sicherlich auch stark vom Unternehmenskontext abhängig.
Bei Web-Agenturen kann ich mir ...“ https://www.linkedin.com/feed/update/urn:li:activity:7485240919311503360/
#WebDev #Vibecoding #AgenticEngineering #KI -
Symbolbild: Wie verändert sich Rollen-Bild (und Selbstverständnis) von Developer:innen? Welchen Zustand will man selber erreichen - schneller sein oder Vollautomatisierung?
("Unemployed" ist natürlich sehr überspitzt)
#WebDev #Vibecoding #AgenticEngineering #KI -
Symbolbild: Wie verändert sich Rollen-Bild (und Selbstverständnis) von Developer:innen? Welchen Zustand will man selber erreichen - schneller sein oder Vollautomatisierung?
("Unemployed" ist natürlich sehr überspitzt)
#WebDev #Vibecoding #AgenticEngineering #KI -
"You can sit anywhere on the spectrum from vibe coding to agentic engineering with the same agent. The thing that decides where you land is verification.
The right spot on the spectrum depends on the stakes. The skill is knowing where to draw the line for each task.
There are two mechanisms. Tests cover the deterministic parts: this input, that output. Evals cover the parts that aren’t deterministic, and the paper splits them in a way I found useful. Output evaluation asks whether the final result is correct. Trajectory evaluation asks whether the path it took to get there, the tool calls and the reasoning, was sound. You want both. An answer that looks right but skipped its checks is more dangerous than one that’s obviously broken.If I had to hand a leader one line from the paper, it’s this: Set the bar at the eval, not the demo. A demo shows an agent can work once. An eval suite with a real rubric shows it works reliably. I keep making this argument; see “Agentic Code Review.”
AI compresses the lifecycle, but unevenly, and the unevenness is the whole story. Implementation drops from weeks to hours. Requirements, architecture, and verification stay slow because they’re judgment work. So specification quality becomes the bottleneck, and verification moves to the middle."
https://www.oreilly.com/radar/the-new-software-lifecycle/
#AI #GenerativeAI #AIAgents #AgenticAI #AgenticEngineering # #VibeCoding #SDLC #SoftwareDevelopment #Programming
-
"You can sit anywhere on the spectrum from vibe coding to agentic engineering with the same agent. The thing that decides where you land is verification.
The right spot on the spectrum depends on the stakes. The skill is knowing where to draw the line for each task.
There are two mechanisms. Tests cover the deterministic parts: this input, that output. Evals cover the parts that aren’t deterministic, and the paper splits them in a way I found useful. Output evaluation asks whether the final result is correct. Trajectory evaluation asks whether the path it took to get there, the tool calls and the reasoning, was sound. You want both. An answer that looks right but skipped its checks is more dangerous than one that’s obviously broken.If I had to hand a leader one line from the paper, it’s this: Set the bar at the eval, not the demo. A demo shows an agent can work once. An eval suite with a real rubric shows it works reliably. I keep making this argument; see “Agentic Code Review.”
AI compresses the lifecycle, but unevenly, and the unevenness is the whole story. Implementation drops from weeks to hours. Requirements, architecture, and verification stay slow because they’re judgment work. So specification quality becomes the bottleneck, and verification moves to the middle."
https://www.oreilly.com/radar/the-new-software-lifecycle/
#AI #GenerativeAI #AIAgents #AgenticAI #AgenticEngineering # #VibeCoding #SDLC #SoftwareDevelopment #Programming
-
The mindset shift every engineer must make to thrive with AI agents: stop writing code, start directing it, and why you won't get lazy doing it. https://hackernoon.com/stop-coding-start-directing-the-paradigm-shift-for-every-software-engineer #agenticengineering
-
The mindset shift every engineer must make to thrive with AI agents: stop writing code, start directing it, and why you won't get lazy doing it. https://hackernoon.com/stop-coding-start-directing-the-paradigm-shift-for-every-software-engineer #agenticengineering
-
It's a bit unsurprising that #ClaudeCode tends to struggle with the same things average software engineers tend to struggle with.
For example database migrations (never edit once committed), and CORS constraints.
-
It's a bit unsurprising that #ClaudeCode tends to struggle with the same things average software engineers tend to struggle with.
For example database migrations (never edit once committed), and CORS constraints.
-
"This Rust rewrite would've taken a team of engineers with full-context on the codebase a year of work. With 1 engineer using Fable & closely monitoring Claude Code, we went from start to 100% of the test suite passing on all platforms in 11 days." by @jarredsumner
-
"This Rust rewrite would've taken a team of engineers with full-context on the codebase a year of work. With 1 engineer using Fable & closely monitoring Claude Code, we went from start to 100% of the test suite passing on all platforms in 11 days." by @jarredsumner
-
No matter who Andrej Karpathy happens to work for at any given time, it has always been worth paying attention to what he says out in public. Here's his latest blog post turned into a summary deck, as part of my regular testing on how well does Microsoft's #PowerPoint agent in #Copilot work with different source formats:
https://slides.jukkan.com/deck/from-vibe-coding-to-agentic-engineering/
-
No matter who Andrej Karpathy happens to work for at any given time, it has always been worth paying attention to what he says out in public. Here's his latest blog post turned into a summary deck, as part of my regular testing on how well does Microsoft's #PowerPoint agent in #Copilot work with different source formats:
https://slides.jukkan.com/deck/from-vibe-coding-to-agentic-engineering/
-
🎧️ AI News #7 mit Fabian Walther und Ole Wendland:
✅️ Noam Shazeer wechselt von Google zu OpenAI
✅️ Fable: vier Tage live, dann von der US-Exportkontrolle gestoppt
✅️ Google besorgt sich 160 Mrd.
✅️ Blasen-Debatte ist zurückJetzt, überall, wo es Podcasts gibt und hier:
🔗 https://www.innoq.com/de/podcast/199-ai-news-7/
🎥 https://www.youtube.com/watch?v=xz_4wvMdLZk -
🎧️ AI News #7 mit Fabian Walther und Ole Wendland:
✅️ Noam Shazeer wechselt von Google zu OpenAI
✅️ Fable: vier Tage live, dann von der US-Exportkontrolle gestoppt
✅️ Google besorgt sich 160 Mrd.
✅️ Blasen-Debatte ist zurückJetzt, überall, wo es Podcasts gibt und hier:
🔗 https://www.innoq.com/de/podcast/199-ai-news-7/
🎥 https://www.youtube.com/watch?v=xz_4wvMdLZk -
@engkiosk Gibt's da schon so etwas wie einen "Industriestandard"? Also beliebtestes Tooling aktuell?
Hatte auch https://opengsd.net/ gesehen, Superpowers im Claude Code Kontext auch
Frage mich generell, wo es jetzt in Web-Agenturen bspw. hingeht: 24/7 mit autonomen Agenten probieren durchrennen zu lassen - oder halt weiterhin eher primär vom Developer:in gesteuert. Ist ja auch Kosten-/Hardwarefrage 🤔
( @workingdraft hat dazu auch Episode veröffentlicht aktuell, #AgenticEngineering vs. #Vibecoding , noch nicht gehört)
-
@engkiosk Gibt's da schon so etwas wie einen "Industriestandard"? Also beliebtestes Tooling aktuell?
Hatte auch https://opengsd.net/ gesehen, Superpowers im Claude Code Kontext auch
Frage mich generell, wo es jetzt in Web-Agenturen bspw. hingeht: 24/7 mit autonomen Agenten probieren durchrennen zu lassen - oder halt weiterhin eher primär vom Developer:in gesteuert. Ist ja auch Kosten-/Hardwarefrage 🤔
( @workingdraft hat dazu auch Episode veröffentlicht aktuell, #AgenticEngineering vs. #Vibecoding , noch nicht gehört)
-
> In practice, the win comes from matching the right model to the right job—planning vs. implementation, small diffs vs. risky refactors, greenfield builds vs. legacy codebases, and quick prototyping vs. production hardening.
-
> In practice, the win comes from matching the right model to the right job—planning vs. implementation, small diffs vs. risky refactors, greenfield builds vs. legacy codebases, and quick prototyping vs. production hardening.
-
AI coding agents make code cheaper to generate, but the real costs move into testing, deployment, operations, and long-term maintenance. https://hackernoon.com/the-hidden-cost-of-agentic-code-generation #agenticengineering
-
AI coding agents make code cheaper to generate, but the real costs move into testing, deployment, operations, and long-term maintenance. https://hackernoon.com/the-hidden-cost-of-agentic-code-generation #agenticengineering
-
Kind of blown away with the #Anthropic #Fable model tonight. It one-shot a hairy problem and the code is solid.
It even found some vestigial code that was never removed but looked like it was still used given the places it was set up.
-
Kind of blown away with the #Anthropic #Fable model tonight. It one-shot a hairy problem and the code is solid.
It even found some vestigial code that was never removed but looked like it was still used given the places it was set up.