Yandex deploys GENA coding agent across a thousand projects
🤖 An autonomous AI coding agent named GENA handles routine tasks across one thousand internal projects at Yandex following its launch in April.
Initially expected to attract only a couple dozen users, the tool expanded rapidly by late summer as engineers assigned it bug fixes and code reviews. Today, 96 percent of company developers regularly use AI agents to automate workflows.
When GENA runs out of memory during a build, it stops and asks engineers to allocate more resources before retrying.
#ArtificialIntelligence ⚡ #CodingAgent ⚡ #SoftwareEngineering
Subscribe
• Sent via TeleFeed
🤖 An autonomous AI coding agent named GENA handles routine tasks across one thousand internal projects at Yandex following its launch in April.
Initially expected to attract only a couple dozen users, the tool expanded rapidly by late summer as engineers assigned it bug fixes and code reviews. Today, 96 percent of company developers regularly use AI agents to automate workflows.
When GENA runs out of memory during a build, it stops and asks engineers to allocate more resources before retrying.
#ArtificialIntelligence ⚡ #CodingAgent ⚡ #SoftwareEngineering
Subscribe
• Sent via TeleFeed
Goldman Sachs restricts AI coding tools to engineers
⚠️ Traders and bankers lost access to advanced developer assistants like Claude Code and GitHub under internal guidelines issued in October at Goldman Sachs. Employees outside engineering spent months building content and monitoring trading flows with the tools before management rolled back access.
Back in July, chief information officer Marco Argenti praised vibe coding by non-technical staff who built quick prototypes before handing them off to engineers.
Spokesperson Nick Carcaterra attributed the decision to productivity and risk management concerns, stating that the experimentation phase has ended.
#GoldmanSachs ⚡ #ArtificialIntelligence ⚡ #SoftwareEngineering
Subscribe
• Sent via TeleFeed
⚠️ Traders and bankers lost access to advanced developer assistants like Claude Code and GitHub under internal guidelines issued in October at Goldman Sachs. Employees outside engineering spent months building content and monitoring trading flows with the tools before management rolled back access.
Back in July, chief information officer Marco Argenti praised vibe coding by non-technical staff who built quick prototypes before handing them off to engineers.
Spokesperson Nick Carcaterra attributed the decision to productivity and risk management concerns, stating that the experimentation phase has ended.
#GoldmanSachs ⚡ #ArtificialIntelligence ⚡ #SoftwareEngineering
Subscribe
• Sent via TeleFeed
Alibaba's coding benchmark tests real-world limits
💻 Alibaba and university researchers launched Co-Design Bench, where top models like GPT-6 Astra complete only 26.4% of complex user change requests successfully.
Unlike older tests where agents hit over 99 percent, this benchmark uses messy real-world tasks that expose deep flaws in multi-step coding.
#ArtificialIntelligence ⚡ #CodingAgents ⚡ #SoftwareEngineering
Subscribe
• Sent via TeleFeed
💻 Alibaba and university researchers launched Co-Design Bench, where top models like GPT-6 Astra complete only 26.4% of complex user change requests successfully.
Unlike older tests where agents hit over 99 percent, this benchmark uses messy real-world tasks that expose deep flaws in multi-step coding.
#ArtificialIntelligence ⚡ #CodingAgents ⚡ #SoftwareEngineering
Subscribe
• Sent via TeleFeed