Interview, Podcast
Embedded AI: The Questions Every CEO is Asking
- Companies are expected to differentiate themselves by embedding AI into workflows to summarize customer conversations, uncover market insights, and serve as an "ear on the ground" for product and strategy decisions.
- Contact centers are predicted to transform by 2028 into "touchless" environments where AI handles data entry and summarization, enabling agents to focus on strategic relationship building and eliminating metrics like average handle time.
- The Net Promoter Score (NPS) for contact center jobs is projected to shift from historically low to positive levels as roles evolve to prioritize mastery, creativity, and proactive customer engagement.
- Developers will utilize AI tools like Hex and Sourcegraph's Cody to generate, edit, and document code, with features including auto-complete, contextual search, and introspection capabilities showing specific source files used for outputs.
- Sourcegraph aims to deploy a viable self-hosted language model for quality Q&A within six months, while offering a pluggable architecture to integrate various model providers including OpenAI, Anthropic, and open-source options like Llama and Alpaca.
- First-mover companies may secure a competitive advantage through data moats built by collecting user preferences and training models over months or years, though competitors may access similar data via API calls.
- The market will see diverse pricing schemes, security postures, and efficacy levels among models, necessitating a combination of language models with informational retrieval engines to maximize AI power.
- Brands are expected to trend toward transparency by disclosing bot interactions and being upfront about accent masking to mitigate reputation risks associated with unexpected AI responses or lack of contextual understanding.
- Data retention practices will increasingly focus on preserving historical information to inform models, despite the ongoing financial costs and risks of maintaining large datasets.
- Privacy and security remain complex, with companies likely facing inquiries regarding data collection and storage practices as they integrate AI into their operations.
- Future AI systems will face challenges regarding "garbage in, garbage out" and hallucinations, requiring right data inputs, fact-checking contexts, and human-in-the-loop verification for code generation.
- Software differentiation will increasingly depend on building workflows that help customers achieve business goals, potentially creating frictionless digital experiences for issues not requiring human intervention.
- AI research will enable companies to identify shifts in customer engagement and reception, serving as a trigger for strategic pivots and allowing agents to provide better deals without long wait times.