消费、配置还是构建？如何真正扩大 AI 应用.pdf

1、Neil SholayVice President AIHow to scale AI adoptionof GenAI pilots are failing*95%*Source:The GenAI Divide:State of AI in Business 2025 by MITs NANDA There is a learning gap for AI tools in most organisations Enterprise data is(still)not AI Ready Integration of AI systems across back-office apps is

2、 lowMost GenAI pilots deliver no measurable impact on P&LWhy?The transformative impact of AI will never be achieved with Horizontal AI Enterprises must explore&master Vertical AI This will lead to growth in Enterprise AI InferencingCo-Pilots&Digital assistants are easy to setup but achieve only marg

3、inal business impactToday focus on horizontal AI with limited business impactFocus on shallow,horizontal AI cases like Co-Pilots Minimal use of AI to transform core business processesDigital assistantBusiness processLess focus oncommodity casesUnlocking high value Vertical use casesTomorrow focus on

4、 vertical AI with P&L impactAI inferencing to solve complex problems in core business processesRe-engineer complete business flow,linking AI models with workflows,business data&domain knowledgeDigital assistantBusiness process5Company runs a fleet of 2000 Electric vehicles,used to install Comms&Sat

5、TV servicesE-Vehicle route planning is more complex than ICE vehicles;Location of vehicle,chargers&equipmentTraffic patterns(real-time)Remaining charge(%)Time to charge battery(mins-hours)Weather conditions during dayLocation,type&duration of jobsCost of energy Kw per hourAvailability of installersA

6、vailability of equipmentAI Inferencing to solve E-Vehicle Route optimisation For AI to be Successful in the Enterprise,We Shall Need6Enterprise data must be accessible&AI readyBusiness users can build AI agents with No code toolsMeasuring b