# Agents@Work: Allen (Autodesk’s AI Workforce Journey)

Source: https://www.youtube.com/watch?v=m3w6buEzjAU
Recap page: https://rapidrecap.app/video/m3w6buEzjAU
Generated: 2026-01-16T00:03:58.767+00:00

---
## Quick Overview

Allen Roh from Autodesk emphasized that the success of their AI agent deployment hinges on establishing clear, data-driven metrics and fostering strong cross-functional collaboration, particularly between the Life Cycle Automation team and Product Marketing, to ensure agents deliver measurable value and align with company goals.

**Key Points:**
- Autodesk views AI agents as the future workforce, noting that tasks taking hours manually can be done in minutes or seconds with agents (00:05-00:20).
- The primary goal established early on was to bring value to everyone in the industry, focusing on processes, products, and films related to design and making (00:55-01:03).
- Success metrics must be clear and focused, specifically tracking data points close to revenue, like conversion rates, rather than broad metrics (01:35-01:46, 07:52-07:55).
- Cross-functional collaboration is crucial; Roh highlighted working closely with Product Marketing, Sales/BD, and Customer Success to align on goals and iterate based on feedback (01:51-02:04, 06:06-06:10).
- The initial expectation was to 'white glove' every initial deployment, but the team quickly realized they needed to move towards scalable, modular systems that still maintain high precision (03:26-03:44).
- A key insight was that AI agents must be trained on core institutional knowledge and best practices, not just generic data, to ensure relevance (07:56-08:19).
- Roh advises leaders starting with AI to start small, focus tightly on a revenue-impacting metric, and avoid assuming the agent will solve everything immediately (08:35-08:40, 17:15-17:22).

![Screenshot at 00:14: Allen Roh, wearing an Autodesk shirt, emphasizes the speed and effectiveness of AI agents, stating that tasks that used to take hours now take minutes or seconds \(00:14-00:20\).](https://ss.rapidrecap.app/screens/m3w6buEzjAU/00-00-14.jpg)

**Context:** The video features an interview segment from the 'Agents@Work' series, hosted by Daniel Vassilev (Founder & Co-CEO of Relevance AI), with Allen Roh, Sr. Marketing Manager for Growth & Activation at Autodesk. They discuss Autodesk's journey in deploying and scaling AI agents within their organization, focusing on setting clear objectives, maintaining data fidelity, and ensuring cross-team alignment for successful adoption.

## Detailed Analysis

Allen Roh outlines Autodesk's enthusiastic yet methodical approach to integrating AI agents into their workflow, starting with the clear goal of delivering value across design, engineering, and construction processes. He stresses that the key to success is not just creating agents, but ensuring they are trained on specific institutional knowledge and clearly defined, revenue-impacting KPIs. Roh details how they initially tried to 'white glove' every deployment but pivoted to a more modular approach to allow for scaling across different products and teams. He highlights the necessity of cross-functional collaboration—involving Product Marketing, Sales/BD, and Customer Success—to gather feedback and iterate on both the agents' performance and the prompts used to guide them. The core philosophy is data-first, confidence second, ensuring that every agent iteration is guided by measurable results, not just excitement about the technology.

### AI Agent Philosophy

- Viewing AI agents as the future workforce; enabling tasks that once took hours to complete in minutes or seconds (00:03-00:20)
- Initial goal was bringing value to design, engineering, and construction (00:55-01:03)
- Agents must be trained on institutional knowledge, not just general data (08:17-08:20).

### Metrics and Focus

- Success requires clear, narrow focus on revenue-impacting KPIs, like conversion rates, rather than broad engagement metrics (01:35-01:46, 07:52-07:55)
- Start small, focus on one key metric, and iterate based on data (08:35-08:40).

### Organizational Implementation

- Success depends on strong cross-functional collaboration across teams like Product Marketing, Sales/BD, and Customer Success (06:06-06:10, 08:47-08:50)
- Initial deployments were hands-on, but scaling required modularity and alignment across different product lines (03:45-04:02, 09:53-10:03).

### Advice for Leaders

- Do not assume initial success is guaranteed; continuous iteration based on real feedback and data is essential (10:22-10:36)
- Ensure everyone is aligned on the goal and process (15:17-15:19).

![Screenshot at 00:24: The 'Agents@Work' title card over a pixelated isometric office illustration, setting the context for the series featuring forward-thinking leaders \(00:23-00:25\).](https://ss.rapidrecap.app/screens/m3w6buEzjAU/00-00-24.jpg)
![Screenshot at 00:38: Title cards identifying the interview participants: Allen Roh \(Autodesk\) and Daniel Vassilev \(Relevance AI\) \(00:38-00:43\).](https://ss.rapidrecap.app/screens/m3w6buEzjAU/00-00-38.jpg)
![Screenshot at 00:31: A screen capture of the visual agent workflow interface being discussed, showing interconnected agents \(00:30-00:32\).](https://ss.rapidrecap.app/screens/m3w6buEzjAU/00-00-31.jpg)
![Screenshot at 02:24: Allen Roh smiles directly at the camera while discussing his initial curiosity about AI agents \(02:24-02:27\).](https://ss.rapidrecap.app/screens/m3w6buEzjAU/00-02-24.jpg)
![Screenshot at 05:05: Daniel Vassilev and Allen Roh walking through the office space, which features a neon sign reading 'HOME OF THE AI WORKFORCE' \(05:04-05:09\).](https://ss.rapidrecap.app/screens/m3w6buEzjAU/00-05-05.jpg)