Celine Xu·5d agoYour Model Isn’t the Moat. Your Deployment Speed IsWhat a self-driving car company taught me about why my best models still take ten weeks to reach a single store
Sarah Marshall·Oct 26Twelve Decisions, One Operating ModelTwelve decisions, made rather than inherited. None of the twelve was about artificial intelligence, and the deployment was never the thing…A response icon1A response icon1
Sarah Marshall·Oct 26What AI Asks of Your People, and What They Need From YouForty-four percent of employees say their organization has started integrating AI. Twenty-two percent say anybody told them the plan. Two…
Sarah Marshall·Oct 26The Five Levels You Are Actually Designing within Your AI DeploymentMost organizations design one of the five levels underneath an AI deployment and inherit the other four. The one they design is the only…
Sarah Marshall·Oct 26What It Costs to Keep AI WorkingTen analysts, ten contracts each, fifty retrievals per contract. That is five thousand retrievals a day, which is also the published daily…
Sarah Marshall·Oct 25Prove AI Works Before You Call It ProductionThree quarters of organizations plan to deploy autonomous agents within a couple of years. One in five reports a mature governance model…
Fana Capital·Oct 25Beyond the Cloud: Why Edge AI is the Next Major Business AdvantageTL;DR: Cloud-based AI is powerful, but it is becoming too expensive, slow, and privacy-invasive for real-time business applications. The…
Sarah Marshall·Oct 25Where AI Operates, and What It Can ReachYour vendor says it can search the web. Which part of the web, whose copy, refreshed how often, and can it reach the contracts in your own…
Sarah Marshall·Oct 25Define the Outcome Before You Compare a Single AI ModelThree vendors demo, the spreadsheet grows, and six weeks later nobody in the room can say in one sentence what the thing is supposed to…
