Latent Space
·
1 day ago
Developers are shifting focus from autonomous agents themselves to the systems that manage them, with "loop engineering" becoming central to how AI agents are controlled and evaluated in production. A survey by Barr Yaron identified coding agents as a key trend, with tools like Claude Code, Cursor, and Gemini CLI replacing traditional IDE-based development by handling broader objectives across multiple files. Enterprises are now deploying AI engineering specialists called forward-deployed engineers to implement AI capabilities, with companies building "software factories" that automate parts of the development lifecycle while keeping humans in control of critical decisions.
The New Stack
·
1 day ago
IBM's second-quarter revenue fell short of expectations at $17.2 billion versus the forecasted $17.86 billion, as enterprise customers redirected spending from software services toward AI hardware like servers and storage. CEO Arvind Krishna acknowledged the company failed to anticipate the magnitude of this shift and adapt quickly enough, causing numerous large deals to slip. Developers will face tighter software budgets and increased pressure to build custom integrations using open-source tools rather than licensed enterprise middleware.
NVIDIA
·
2 days ago
NVIDIA Nemotron Labs promotes open AI models that enterprises can customize and control for domain-specific tasks, contrasting with closed proprietary models. Companies like Harvey achieved legal task accuracy matching frontier models at 10x lower cost, while Arcee AI reached inference costs of approximately 90 cents per million tokens, roughly 20x cheaper than comparable closed models. This shift enables organizations to build specialized AI applications tailored to their specific workflows and data rather than adapting their needs to existing general-purpose models.
The New Stack
·
2 days ago
Valantor acquired document intelligence company EyeLevel and launched its Enterprise Visual Intelligence platform to process unstructured data like PDFs, handwritten documents, and complex forms that large language models struggle to access. The company's GroundX platform processes visually complex documents through multi-pass orchestration, achieving 96% accuracy on policy questions for Air France-KLM and automating 85% of customer inquiries for AskVet. Organizations can now access the approximately 80% of corporate knowledge trapped in unstructured documents while maintaining data sovereignty and reducing processing costs by decomposing documents into smaller elements for cheaper AI models.
The Neuron
·
2 days ago
UnitPay launched a no-code billing platform designed to help AI product companies track usage-based revenue and automatically handle invoicing and payments. The platform supports setup in 10 minutes with a simple SDK integration and processes 50 million events per month with sub-12 millisecond latency. Customers gain visibility into per-model margins, churn prediction 30 days in advance, and can adjust pricing models without code changes.
