# The Moment You Can Clone Your Best Engineer | Daniel Siegel | TEDxHochschuleBremerhaven Source: https://www.youtube.com/watch?v=tXymCjB25R8 Recap page: https://rapidrecap.app/video/tXymCjB25R8 Generated: 2026-01-09T22:02:57.186+00:00 --- ## Quick Overview The ability to clone the best engineer's output by integrating AI agents with existing engineering tools is becoming a reality, exemplified by the rapid adoption curve of AI technologies compared to the internet, suggesting that the future of engineering will involve AI systems that estimate, generate, and execute tasks, potentially leading to an 84% success rate in Multi-Agent Systems by 2025. **Key Points:** - The speaker argues that the future of engineering involves close collaboration with AI, specifically working with Artificial Intelligence agents that Estimate, Generate, and Control & Execute tasks. - The adoption rate of AI (like ChatGPT reaching 100 million users in about 4 months) significantly outpaces the internet's adoption rate (which took 13 years to reach 800 million users), indicating accelerating technological evolution. - The speaker cites the 'Attention Is All You Need' paper from 2017 as a critical inflection point that laid the foundation for modern Large Language Models (LLMs). - The current challenge is integrating these powerful LLMs with engineering tools (like CAD software) which are often 30 years old, creating a gap that needs to be bridged. - The speaker projects that by 2025, Multi-Agent Systems (MAS) will achieve an 84% success rate in engineering tasks, up from an estimated 40% success for initial agents in 2023. - The key to this acceleration is developing a new modality of understanding geometry and creating protocols that allow specialized AI agents to communicate and determine the next steps for complex engineering tasks. ![Screenshot at 11:20: The speaker projects an exponential growth curve for AI capability \('Skillset Score\*'\) from 2022 \(ChatGPT, ~30% score\) through 2025 \(Multi-Agent Systems, ~84% success\), contrasting it with the slower adoption of the internet.](https://ss.rapidrecap.app/screens/tXymCjB25R8/00-11-20.jpg) **Context:** The presentation takes place at a TEDx event hosted by Hochschule Bremerhaven, featuring a speaker named Daniel who discusses the future of engineering in the context of rapidly advancing Artificial Intelligence, particularly Large Language Models (LLMs) and AI agents. The stage design incorporates nautical themes, including large octopus tentacles and fishing nets, possibly referencing Bremerhaven's maritime history or the complexity of deep-sea engineering challenges. ## Detailed Analysis Daniel Siegel presents a vision for the future of engineering where AI agents, built on LLMs, become indispensable digital coworkers. He begins by highlighting the historical inflection points in engineering evolution (1950s drafting boards, 1990s CAD, and 'Now' incorporating AI agents) and emphasizes the break-through nature of the 2017 'Attention Is All You Need' paper that enabled modern LLMs. He contrasts the rapid adoption of ChatGPT, achieving massive user numbers in months, with the internet's decade-plus growth, suggesting AI's acceleration is far steeper. The core problem identified is that while LLMs are brilliant at language, they often lack the specialized, geometric understanding needed for engineering tasks, especially when interacting with legacy tools. Siegel proposes that the quality of the resulting system (Qa) depends on four weighted factors: Quality of Agents (Qs), Quality of Model (Qm), Quality of Context/Prompt (Qc), and Quality of Tool (Qt). To realize the future potential, engineers must focus on increasing the quality of all these components, particularly by developing multi-agent systems where specialized agents coordinate their efforts to solve complex problems, aiming for an 84% success rate by 2025. ### Introduction to AI in Engineering - Discusses the future of engineering requiring collaboration with AI agents that Estimate, Generate, and Control & Execute - Mentions the 2017 'Attention Is All You Need' paper as the foundational inflection point - States that current engineering tools are outdated compared to new AI capabilities. ### The Acceleration of AI Adoption - Compares the adoption rate of ChatGPT (1M users in <5 days, ~30% skillset score in 2022) against the internet's 13-year timeline to reach 800M users, showing AI's exponential speed. ### Framework for Agent Quality (Qa) - Presents the formula Qa = f(w1Qs, w2Qc, w3Qt) where Qa is the overall quality dependent on Agents, Model, Context (Prompt), and Tool quality - Notes that the model quality improved from ~30% (2022) to ~40% (Agents in 2023) and is projected to reach ~70% (2024) and 84% (Multi-Agent Systems in 2025). ### The Engineering Challenge - Highlights that while LLMs understand text well, they struggle with understanding geometry, which is crucial for engineering tasks like designing off-road vehicle wheels - Shows simulation results where the selected wheel design meets all necessary mechanical properties. ### The Future - Focuses on the necessity of Multi-Agent Systems where specialized agents communicate and use tools to solve complex engineering problems, moving beyond simple auto-completion to actual complex task execution. ![Screenshot at 00:15: Speaker introducing the topic with an animation of a projectile, likely symbolizing speed or impact.](https://ss.rapidrecap.app/screens/tXymCjB25R8/00-00-15.jpg) ![Screenshot at 00:26: Speaker interacting with an assistant, discussing the need to prepare a design space for a specific task.](https://ss.rapidrecap.app/screens/tXymCjB25R8/00-00-26.jpg) ![Screenshot at 00:32: Digital interface showing a 3D model of a car wheel being initialized, with status updates like "\[READY\] Workspace ready."](https://ss.rapidrecap.app/screens/tXymCjB25R8/00-00-32.jpg) ![Screenshot at 00:54: The screen displays 9 design options for a 20-inch off-road wheel, which the assistant prepared.](https://ss.rapidrecap.app/screens/tXymCjB25R8/00-00-54.jpg) ![Screenshot at 02:04: Simulation results screen showing a color-coded stress analysis \(red indicating highest stress\) on the selected wheel design, with simulation checks marked as complete.](https://ss.rapidrecap.app/screens/tXymCjB25R8/00-02-04.jpg)