Blog
Notes on applied AI.
Field notes on applied AI. A running log of what we’re reading, building and thinking about — the releases, the experiments, the deployment gotchas. Auto-curated daily, with notes on what’s shipping, what’s hype, and what’s actually worth your attention. From the IT Consulting AI team.
Latest
-
Why AI Diagnoses Outperform Doctors Yet Risk Rapid Obsolescence
Specialized AI systems now match doctors in diagnosis but risk quick obsolescence without constant model updates and real-world adaptation.
-
Formal Verification in AI: Betting on Reliability over Hype
Pramaana Labs raises $27M to apply formal verification to AI, focusing on high-risk fields where reliability is critical and errors are costly.
-
Salesforce’s Fin Acquisition Signals Deeper AI Agent Integration
Salesforce’s $3.6B acquisition of Fin deepens AI agent integration in its platform, signalling a shift in enterprise AI and vendor lock-in risks.
-
Why Microsoft’s Mirage Misses the Mark on Object Tracking for Video AI
Microsoft’s Mirage reduces compute needs in video AI by using latent space memory but fails at reliable moving object tracking across scenes.
-
AI-Driven Influence Campaigns Reveal New Security Risks for Tech
Google and OpenAI uncover AI-powered Chinese influence campaigns, exposing evolving security risks that demand a new strategic response from tech leaders.
-
AI as a Strategic Weapon: The New Cold War Playbook Unveiled
Anthropic’s essay reframes AI as a strategic weapon in a geopolitical Cold War, calling for binding audits and global governance to manage risks.
-
SpaceX’s IPO Value Hinges on Ambitious Space Data Center Plans
SpaceX’s IPO value largely depends on its speculative plans for space-based data centers, marking a bold bet on future cloud infrastructure beyond Earth.
-
Public Listings Will Drive More AI Token Price Increases
The public listing of major AI companies will drive up token prices, forcing businesses to prepare for higher costs and more concentrated vendor control.
-
OpenAI Declares Chat Dead to Push Autonomous Agent Superapp
OpenAI plans to transform ChatGPT into an autonomous agent superapp, raising concerns about vendor lock-in, integration complexity, and loss of software flexibility.
-
Microsoft’s Data Claims Don’t Hold Up Under Detailed Scrutiny
Microsoft claims its MAI models use only licensed, clean data, but closer analysis reveals reliance on unlicensed web sources similar to other AI labs.