Practical AI 2025 was the year of agents, what's coming in 2026?
340 snips
Jan 9, 2026 A whirlwind recap of AI's evolution in 2025 sets the stage for the discussions on the impactful rise of AI agents. Hear about the varied success of these agents and how they’ve accelerated productivity. The conversation dives into the complexities of multimodal AI, from video to text, and the urgent need for better orchestration skills. Chris and Daniel tackle the challenges of infrastructure and energy limits while forecasting a new era of consumer AI with affordable hardware. Predictions abound for 2026's AI landscape, hinting at exciting developments ahead.
AI Snips
Chapters
Transcript
Episode notes
Pair Domain Expertise With Prompt Iteration
- Invest in domain expertise and learn to craft and iterate prompts over many steps to reach production quality.
- Connect the right data sources and tools; lacking that integration often causes agent projects to fail.
Multimodal Input Outpaces Output
- Multimodal inputs (text, image, audio, video) are common in business, but multimodal outputs remain mainly text or speech.
- Richer multimodal output experiences combined with agents are a clear near-term opportunity.
Reasoning Models Add Latency And Cost
- Reasoning models stream chain-of-thought tokens that mimic reasoning but add significant latency and compute cost.
- They enable dynamic orchestration but raise power and inference-cost tradeoffs for production systems.