Inside the Agency Model: Pressure, Performance & the Power of Manus AI Part 3

In this brutally honest episode, Danny McMillan and Dorian dissect the realities of working in and with agencies. From client expectations to internal pressure, Danny opens with a stark insight:

“You're taking on their pressure too — to turn the ship.”

He reflects on how experience helps determine what truly works and what’s just “proven strategy nonsense.”

The Pitfalls of the Big Agency Portfolio

Dorian highlights a harsh truth about agency structure:

“You’ll get the lowest experience manager assigned to your account. It just doesn’t matter to them.”

Clients are often drawn in by flashy portfolios, but in reality, smaller accounts get junior staff and rarely the attention of top talent.

Danny reinforces this, noting how agencies scale through SOPs and “sweatshop workers,” saying:

“Two out of ten account managers might be elite — the rest are filler.”

Software + Managed Services = Locked-In Mediocrity

Software companies that bundle software with account management often fall short. Danny explains that many account managers are “demo people” turned strategists, lacking deep PPC knowledge or cross-platform experience.

“They run out of strategies by month two.”

The root issue? The system trains people for one tool, not the dynamics of diverse, real-world businesses.

AI & Automation: The Accountant Case Study

Shifting gears, Danny shares how automation is reshaping his internal team. Manual tasks are disappearing, he reallocates those team members to high-leverage roles like data validation, dashboard reporting, and performance benchmarking.

“Efficiency shouldn’t equal job loss — it should mean more meaningful output.”

 

Manus AI: Autonomous Power Tool or Overhyped Wrapper?

One of the most compelling parts of the episode is Danny's deep dive into Manus AI. While it runs on Claude, he insists it’s far from just a wrapper:

“Manus is like dealing with a human. It uses 30 native tools, runs its own checklists, and breaks down tasks autonomously.”

Dorian explains how Manus scrapes, previews, and clicks through websites, operating with depth beyond ChatGPT. Danny recounts solving major business problems — from venue bookings to accounting automation — all in four days for just $240 in credits, declaring:

“The ROI was absolutely fucking priceless.”

Tool Talk: LM Studio vs. Manus

Danny experiments with LM Studio, an offline local LLM platform. While it's promising for airplane-mode workflows, he admits it’s clunky compared to Manus. His final take?

“Do you want to fiddle with open source to save a few pennies or use something that just works?”