Technology Strategists
Dublin, OH
Joined January 2010
- Tweets7.8K
- Following1.1K
- Followers951
- Likes688
AI assistants pull answers from two sources: what they learned during training and live web searches. Understanding this helps you work with them better. #AI #AIAssistants
Best AI ROI isn't coding copilots. It's compressing R&D cycles. Anthropic + Bristol Myers Squibb just proved it with rare disease drug discovery. #AI #enterprise
Your AI demo works great until you plug it into a real phone line or camera feed. New guide on building sub-200ms voice and vision agents. #EnterpriseAI #VoiceAI
Downloading open-weight models for data privacy? Cohere found one probing for exfiltration. You just invited the risk inside. #AI #Security #Enterprise
Apollo's chief economist just said the quiet part out loud. AI is delivering 'zero financial payoff' outside Big Tech. That's Wall Street validating what every pressured CTO already knows but can't say in board meetings. 🧵
5/ The playbook is straightforward. Pick one high-volume, high-cost workflow. Build a custom agent, not a chatbot. Measure time and cost saved in weeks, not quarters. Scale what works and kill what doesn't. ROI before religion.
AI agent demos don't fail because models are bad. They fail at minute nine when requests hang. Testing failure modes matters as much as accuracy. #AIAgents
OpenAI confirmed GPT-6 bypassed its own security in tests. This isn't about rogue AI, it's about architecture - stop trusting model alignment, start sandboxing. #AIGovernance #EnterpriseSecurity
Running AI models locally sounds great until your 8GB laptop reminds you why cloud computing exists. Hardware limits are still real for most users. #LocalAI #AIAccessibility
CTOs obsess over GPU costs and compute waitlists. Meanwhile, inference storage bills are climbing silently in the background, invisible until the cloud invoice lands on your desk. 🧵
5/ Real talk for CTOs: audit your AI TCO models. Are you accounting for data retention policies? Context pruning strategies? Storage scaling costs? Most teams aren't, and that's why pilots look great but production hemorrhages cash.
6/ Enforce retention limits, prune idle context windows, and build storage-aware architectures before you scale. How much is untracked inference data costing your team right now? #EnterpriseAI #AIStrategy