AI Weekly #1: Enterprise Focus

Headlines and my take

Gartner: releases a report with some bold claims about Agentic AI, including 40% of enterprise applications will be integrated with task-specific AI agents by 2026, and CIOs have a crucial three- to six-month window to define their agentic AI strategy. I unpack here what is real and what is hype, and what leaders should do for now.

Workday: Q2 revenue is up 12.6% YoY. They attribute the strong performance to their artificial intelligence (AI) initiatives. “More than 70% of our core customers are now leveraging Workday Illuminate,” from the prepared remarks. Enterprise leaders elsewhere are quietly taking notes, and it deserves a deeper look to the fiancial reports to understand their success.

Apple: opensources FastVLM on Hugging Face with 0.5, 1.5 and 7B real-time vision-language models (VLMs). Apple always commits to privacy, that is why they develop small models that can run on device and never transfer data to a server. As text LLM progress shows signs of slowing down, AI labs will pay more attention to multimodal/vision. If your enterprise handles images, videos, audio, especially with regulatory constraints, now is time to start experimenting with multimodal LLMs. More detail from Apple ML: FastVLM research.

Meta: enters a $10B+ six-year deal with Google Cloud to use its infrastructure, including TPUs, to power Meta LLM needs. Two months ago OpenAI also began testing Google’s TPUs. I analyzed before that having its own chips make Google has a clear advantage in the AI arms race while other competitors can only dream of, and there is also a big demand for inference chips up to grab.

MIT: builds a deep learning system, VaxSeer, that uses a protein language model to predict dominant flu strains and pick vaccine candidates to improve accuracy in seasonal strain selection. They use small transformer language models over protein sequences; no general-purpose LLM is involved. Better strain picks can reduce severe cases and improve public health. As health becomes a bigger priority in everyone’s life, keeping an eye on AI developments in health care can be a smart move for retail investors.

Enterprise AI deployments

You can get some inspiration from what enterprises are releasing. Beware, real impact mixes with press hype.

Citi: builds “Advisor Insights” and “AskWealth”, an AI dashboard plus GenAI assistant that show timely market and portfolio messages and give staff instant answers, to speed up client communications and service. Citi newsroom

Salesforce: builds an enterprise simulation environment with “CRMArena-Pro” and an agentic benchmark so that teams can test and measure agents on real business workflows before go live. Salesforce AI Research

Cleveland Clinic: rolls out Dyania Health’s “Synapsis AI”, medically trained LLMs that automate chart review to find eligible trial participants, to speed up clinical trial recruitment. Cleveland Clinic newsroom

Singapore: gives agencies access to Gemini on Google Distributed Cloud in a secure setup and an AI agent sandbox, to test and refine agents for public services with tighter security. GovInsider Google Cloud blog

FairPrice Group: deploys agentic AI assistants built on Gemini, Vertex AI and related tools, including multimodal smart cart helpers, to guide shoppers and help store and workforce tasks. CDOTrends Google Cloud

Broadcom / VMware: launchs “VMware Tanzu Data Intelligence”, an AI-ready lakehouse that unifies multimodal data so teams can build analytics, applications, and agentic AI at scale. Broadcom news

Cloudflare: releases CASB integrations that scan ChatGPT, Claude and Gemini for misconfigurations, sensitive data exposure and compliance risks, to help companies secure employee AI use. Cloudflare blog Cloudflare press