engblogs

MorphLLM revolutionizes AI code editing by enabling fast, precise, and semantic single-line changes, vastly improving agent efficiency and developer productivity.

OpenAI and Retro Biosciences leveraged GPT-4b micro to engineer enhanced Yamanaka factor variants, achieving a 50x increase in stem cell reprogramming efficiency and demonstrating accelerated rejuvenation potential through AI-driven protein design.

Pinterest redefined developer experience by building PinConsole, an Internal Developer Platform on Backstage, to unify tooling, reduce complexity, and scale engineering productivity across its vast ecosystem.

Blue J leverages GPT-4.1 and a proprietary Retrieval-Augmented Generation system to deliver expert, accurate, and scalable AI-powered tax research across multiple countries, driving rapid iteration, trust, and efficiency in complex, regulated domains.

Instacart leverages AI-driven development to accelerate engineering workflows, enhance productivity, and foster a culture of AI-assisted innovation through strategic adoption, specialized tooling, and structured learning programs.

Modular Platform 25.5 advances scalable AI deployments with Large Scale Batch Inference, enhanced integrations, and growing community innovations including Mojo's rapid adoption and GPU-focused projects.

MIXI rapidly deployed ChatGPT Enterprise to enhance operational efficiency, foster AI-driven innovation, and transform communication across its diverse business divisions with significant time savings and user adoption.

MIT researchers developed FastSolv, a machine learning model that significantly improves prediction of molecular solubility in organic solvents, enhancing drug synthesis and environmentally safer solvent selection.

Dropbox’s AI adoption journey highlights strategic leadership, tailored tooling, and a culture of collaboration to achieve over 90% AI integration across engineering workflows, transforming productivity and software development.

Mariana Garavaglia, DoorDash’s CPO, discusses how AI is transforming workforce productivity, literacy, and personalized employee development within the company's operator culture.

MIT researchers employed sparse autoencoders and AI analysis to interpret the inner mechanisms of protein language models, revealing specific protein features they encode and enhancing model explainability for biological applications.

Effective trust calibration in AI software requires aligning user trust with system capabilities through adaptive mechanisms, timing strategies, transparency balance, and clear distinction between cooperative and delegative systems to enhance safety and usability.