Opening remarks

Embrace the AI Era: Dr. Jon Krohn’s Insights on Agentic AI and Strategic Adaptation

(This article was generated with AI and it’s based on a AI-generated transcription of a real talk on stage. While we strive for accuracy, we encourage readers to verify important information.)

Dr. Jon Krohn, MC at the Web Summit Vancouver 2026 AI Summit, welcomed attendees, setting the stage for an afternoon dedicated to artificial intelligence, particularly agentic AI. He introduced himself as the host of the Super Data Science podcast, author of “Deep Learning Illustrated,” and an AI PhD from Oxford. Dr. Krohn also lectures at Columbia and NYU, supervises research, and is CEO of Y Carrot, an AI software consultancy.

He highlighted AI’s rapid acceleration, comparing the current moment to February 2020, just before the profound societal shifts of the COVID-19 pandemic. Referencing the “Something Big is Happening” blog post, Dr. Krohn stressed that AI’s imminent arrival will drastically reshape life, even if its full impact is hard to envision today.

AI has swiftly surpassed animal and less intelligent human capabilities. Currently, AI operates between less intelligent humans and Einstein-level intellect. Within the next year, it is projected to discover new physics and exceed Einstein across various white-collar tasks, potentially leading to artificial superintelligence and the “singularity.”

A chart demonstrated AI’s progression: GPT-2 (2019) handled tasks taking seconds. By 2022, GPT-3.5 achieved 50% accuracy on tasks lasting 30 seconds to a minute. GPT-4 then tackled tasks requiring several minutes. Recent releases like 01, GPT-5, and Cloud Opus 4.6 now competently manage tasks taking humans several hours, including complex 14-hour programming and machine learning assignments, showing an alarming linear acceleration.

Agentic AI frameworks are a key driver of these advancements. Dr. Krohn defined an AI agent as an LLM agent that runs tools in a loop to achieve a goal. These modern LLM agents are uniquely powerful due to their ability to interpret complex commands and context, distinguishing them from earlier agents.

These agents can access a vast array of tools, including web search, file systems, and Google Maps, facilitated by frameworks like Model Context Protocol (MCP), which connects to millions of tools from thousands of open-source providers. Agents work iteratively, often with master agents delegating goals to hundreds of sub-agents for a single task. Dr. Krohn is authoring “Engineering AI Agents” on this subject.

Despite increasing automation, Dr. Krohn assured that AI careers are not doomed. Data shows a flourishing demand for AI professionals and roles augmented by AI, such as nurse practitioners, paralegals, and software developers. Individuals can thrive by collaborating with AI, augmenting their capabilities, and investing in hard-to-automate skills like judgment, stakeholder communication, and real-world domain expertise.

He urged continuous curiosity and engagement with leading AI tools, including generative AI conversational interfaces (Cloud, Gemini, ChatGPT), code-generation tools, and personal agentic tools like Cloud Co-work, Gemini agents, and Open Claw. Utilizing these tools is essential for staying current and competitive in the evolving landscape.

For organizations, Dr. Krohn addressed the challenge of “secret cyborgs”—employees using external AI tools without company knowledge—offering solutions discussed in his podcast with Professor Ethan Molich. To ensure AI projects yield a return on investment, a rare feat for only 5% of initiatives, he advised focusing on execution metrics over adoption metrics.

Larissa Schneider of Unframe, a podcast guest, highlighted that the percentage of a workflow automated directly correlates with the value captured, emphasizing efficiency over mere usage. Dr. Krohn also mentioned his role as editor for a Pearson book series, including “Building Agentic AI” and “Becoming an AI Orchestrator,” and a comprehensive AI training video on his YouTube channel.

The upcoming session will explore the power shift from training to inference in the AI industry, examining AI infrastructure, competition, and collaboration in the “new diplomacy of the AI chip wars” with Rodrigo Liang and Alexander Pigman.