The Architecture of Agency: Lessons from Autodesk CEO Andrew Anagnost

Autodesk CEO Andrew Anagnost argues that long-term success does not come from strategy alone. Instead, it requires a disciplined balance of courage, scholarship, and empathy. Many leaders fall into a cycle of "thrashing," where they waste energy by pandering to extremes and avoiding hard truths. Anagnost suggests that by cutting stagnant projects and hiring mission-driven people rather than mercenaries, leaders can build a durable competitive advantage. This approach is useful for executives and founders who need to understand why making painful decisions now is the only way to scale effectively later.

The Hidden Cost of Thrashing

Many leaders try to please every internal group, which creates an illusion of progress while wasting time and money. Anagnost calls this "thrashing." When a company lacks a clear mission, it gets stuck between competing interests, such as the old guard and new rebels. This leads to a stalemate where no real strategic decisions happen.

The solution is not to force consensus, but to build a leadership style that refuses to reward political games. By clearly explaining the reason for every difficult change, a leader can connect different factions. This takes time, and it requires the patience to handle the pushback that comes with cutting favorite projects.

"If you really wanna be effective you have to prevent thrashing. And when you pander to one extreme or the other in a company, in a society or anything, you thrash and you thrash back and forth and you had this illusion of progress but you don't."

-- Andrew Anagnost

Why the Obvious Fix Makes Things Worse

Conventional wisdom says that to scale, you should keep all projects running to maximize innovation. Anagnost disagrees, arguing that failing to prune a portfolio is a major risk. His process for killing projects is simple: does the company have a right to win here, and how far is the project from the core business?

Keeping weak projects alive dilutes resources and prevents investment in back office infrastructure. This foundation is unglamorous but necessary for scaling. Anagnost’s decision to shift investment to these systems early in his tenure was unpopular, but it was required to move the company to a subscription model.

The 18-Month Payoff of Vertical AI

While many companies are rushing to use horizontal AI tools, Anagnost warns of a hidden trap: the cost of compute. Using a massive, general model to solve specific, vertical problems leads to high costs and low precision.

The real advantage comes from building custom models trained on proprietary data. This creates a moat that horizontal competitors cannot easily cross. Over the next 18 months, companies that focus on precision rather than prompts will likely see lower costs and more reliable workflows.

"It's expensive to use a horizontal tool to solve vertical problems. The ability to deliver precision, speed and total cost of ownership is going to define vertical software in the future with AI."

-- Andrew Anagnost

Where Immediate Pain Creates Lasting Moats

Anagnost argues that empathy is a strategic necessity, not a soft skill. Courage without empathy is just meanness, which eventually causes the organization to turn against the leader. By practicing empathy, which he defines as understanding the impact of one's actions, a leader can make difficult decisions like layoffs or cancellations while keeping the team's trust. This creates a durable advantage because the team stays aligned even when the world changes.

"Empathy is the way you glue courage with an outcome. And I think anybody that thinks it's other than that, is just confused."

-- Andrew Anagnost

Key Action Items

• Audit your projects: Evaluate your work against the right to win and adjacency tests. If a project is too far from your core, cut it. Time horizon: Immediate.
• Prioritize the back office: Identify the internal systems that prevent you from scaling. Invest here even if it feels like non-innovative work. Time horizon: Next 6-12 months.
• Filter for mission-driven talent: Stop chasing mercenaries who want top pay for short-term gains. Focus on hiring people who love the problem you are solving. Time horizon: Ongoing.
• Develop custom AI models: Move away from relying only on horizontal AI tools. Build custom models on your own data to gain precision and cost advantages. Time horizon: 12-18 months.
• Practice scholarship before action: Resist the urge to be impulsive. Do the research before making major pivots. Time horizon: Ongoing.
• Communicate the why relentlessly: When making unpopular decisions, over-communicate the reasoning. This builds the trust needed to survive the thrashing phase of a turnaround. Time horizon: First 12-24 months of leadership.