Google's flagship Gemini 3.5 Pro remains months behind its original schedule after the model fell short of internal expectations on coding and complex reasoning tasks; the company reset its training data in June to address the gap but results were reportedly still disappointing.
why it mattersGoogle is now the one flagship lab visibly behind on coding capability, just as Anthropic, OpenAI and open-weight Chinese labs all ship competitive coding models.
Bloomberg
The July release of Visual Studio Code adds a dedicated background process for running agent sessions that multiple windows can connect to, a new experimental editor panel for reviewing agent-generated diffs, the ability to run terminal commands from chat with a "!" prefix, and a preview of a redesigned workbench UI.
why it mattersIt's Microsoft further restructuring the editor itself around long-running, multi-window agent sessions rather than single inline suggestions.
Visual Studio Code
Emergent raised a $130 million Series C led by Creaegis at a $1.5 billion valuation, a five-fold jump in six months; the AI-agent app-building platform now has an annualized run rate of $120 million and more than 200,000 paying customers.
why it mattersIt's one of the fastest unicorn runs of the year, and a sign investor appetite for AI app-building tools hasn't cooled despite a crowded field.
TechCrunch