A Million saved is a Million earned - ROI on AI

Large Internet Provider saves millions with RAG, who's measuring AI ROI and a 1 hour ROI deep dive

Subject: A million saved is a million earned - ROI on AI

Welcome to the Executive AI Playbook! We highlight successes, cautionary tails and deep dives into Strategic AI stories and playbooks. We focus on actionable blueprints from industry for your Gen AI Strategy.

Denys Linkov, Enterprise AI @ Voiceflow.

Successful launches - Comcast saves millions with generative search

Earlier this month, Comcast released one of the strongest public case studies of using Generative AI with their AMA (Ask Me Anything) assistant for customer support agents. They reduced the search time for support agents by 10% with an 80% satisfaction rate from agents. This time reduction led to millions in reported savings. The short paper provides a blueprint of experiments and implementation details for teams to follow for implementing their own RAG search. Dive deeper into the case study here.

Cautionary Tales - 60% of leaders are not measuring AI ROI precisely

In contrast to the Comcast case study, A16z surveyed 70 Enterprise AI leaders and found that 60% were not measuring AI ROI precisely or at all. As the AI space continues to mature, CEOs will continue to ask hard questions, with 58% expecting product improvements. Without measuring ROI many teams will struggle to plan FY 2025 projects that show meaningful business results.

Deep dive - AI ROI From startup to enterprise

Last week were were joined by Chip Huyen (Startup exit, VP OSS at Voltron, Author Designing ML Systems) and Colin Guilfoyle (SVP of CX at Trilogy) to discuss measuring the ROI of AI from a startup and enterprise lens. This candid conversation covers important topics from leaders in the space ranging from GPU cloud spend optimization to scaling across lines of businesses.

Chip shared her story about:

  • Budgeting for a complex 2 year technical build

  • Tradeoffs in GPU spend vs Engineer spend

  • Difference in AI finance in a mid size startup

Colin shared his story of:

  • Building an AI practice from scratch for customer support

  • Structuring the team for success

  • Onboarding 60+ product lines

  • Automated 60% of L1 support tickets ($105k in Q4)

Watch the full hour discussion here

Questions to ask your team
How are we measuring success in our next Gen AI launch?

Until next week,
Denys - Enterprise AI at Voiceflow

P.S Find the read helpful? Forward it to a friend working on AI strategy.