Natalia Quintero, Head of Consulting at Every, shares AI adoption success patterns gleaned from consulting with over 100 companies over the past year, emphasizing the importance of leadership initiative and internal champion development. She shares a case study of a private equity firm that reduced investment memo creation from 3 weeks to 30 minutes, and a framework for maximizing engineering team productivity. Most notably, she unveils 'Claudie,' an AI project manager she built through 'vibe coding,' demonstrating how it reduced 15 hours of weekly administrative work and enabled deeper focus on client communication.

1. Two Key Elements for Successful AI Adoption

In 2026, Natalia has discussed AI adoption with over 100 companies over the past year. Among the many companies she observed, there was a clear difference between those that successfully leveraged AI and those that didn't. Natalia emphasizes that the key isn't simply adopting expensive tools but rather 'organizational effort.'

For AI to be useful in a company, it requires organizational effort. (...) For AI to become a high-leverage tool in a company, it needs to come top-down. It doesn't work like the old way of software adoption where the CTO buys it and hopes employees figure it out.

Two elements are needed for successful AI adoption:

1. Top-down approach: Leadership must understand that AI is a powerful tool that fundamentally changes how work is done, and seize and manage opportunities at the organizational level. The most powerful motivation comes from the CEO personally using and experimenting with AI.
2. Bottom-up champion development: Employees must be empowered to work creatively with AI. It's important to discover and support 'AI champions' who experiment without fear of failure and spread AI usage to colleagues.

2. From 3 Weeks to 30 Minutes: A Private Equity Firm's Innovation

Natalia introduces an amazing case from a private equity (PE) client. Jonathan, a partner at the firm, was both technically skilled and deeply understood organizational dynamics. He mapped every task performed by all investors down to the finest detail.

We looked at the company's long list of tasks and marked where AI could provide really high leverage. (...) It was possible because we understood very specifically not just how they describe their workflows, but how they think about and approach their work.

Thanks to this detailed analysis, they were able to revolutionize the grueling task of writing 'Investment Memos.' Previously, analysts needed 2-3 weeks to synthesize the company's extensive investment arguments and data into memos. But by connecting the company's internal knowledge (IP) and context to AI with carefully crafted prompts reflecting their thinking style, they can now get high-quality drafts in 30 minutes.

3. The 'Plan-Delegate-Assess-Compound' Framework for Engineering Teams

Natalia also shares an interesting pattern discovered while working with engineering organizations at tech companies. She says engineers need the following 4-step framework to effectively leverage AI:

1. Plan
2. Delegate
3. Assess
4. Compound

Many engineering teams are proficient at delegating tasks to AI and assessing results, but surprisingly often skip the 'Planning' step. Without proper planning, AI can only handle simple tasks and cannot solve complex, large problems.

They weren't getting far because they had no planning stage. (...) Once they realize how important proper planning is, you consistently see engineers completing 2 weeks of work in a single afternoon.

4. A Non-Developer's Revolution: 'Vibe Coding' at 6 AM

Natalia confesses that while she doesn't have a technical background, she's recently become deeply immersed in 'vibe coding' — the practice of intuitively writing code by conversing with AI (primarily Claude) without professional programming knowledge.

She felt that during the '9-to-5' workday, meetings and immediate tasks left no time for experimenting with new tools. So she and her colleague Natash dedicated the hours from 6 AM to 9 AM to building an AI project manager.

To be honest, I'm now a complete 'vibe coding addict.' (...) We had to scrap and restart the project manager agent three times. But ultimately, we were able to build a system that saves us over 14 hours per week.

5. Introducing AI Project Manager 'Claudie'

Natalia demonstrates 'Claudie,' the AI agent she built herself. Claudie handles the complex management of consulting projects.

5.1 Claudie's Structure

Claudie is hosted on GitHub and structured as follows:

• claude.md: A file like Claudie's 'job description' — specifying who it is, who it works with, where it gets data, and what principles (data accuracy, proactive responses, etc.) it must follow.
• MCPs (Model Context Protocols): Pipelines connecting to external data sources like Gmail, Google Calendar, Google Drive, and meeting transcripts.
• Skills & Tasks: Command sets for performing specific tasks (e.g., new client setup, weekly updates).

5.2 Live Demo: New Client Onboarding

Natalia opens her terminal (Warp) and gives Claudie the new client setup command. Claudie then performs:

1. Information gathering: Independently searches the user's email, calendar, drive, and meeting records to understand all context about the client.
2. Sub-agent execution: Simultaneously runs 4 sub-agents to scrape information from different sources.
3. Database update: Automatically fills in the project management spreadsheet (dashboard) based on collected information, assigning unique IDs to people and tasks for systematic database-like management.

6. Conclusion: Freed from Spreadsheets, Focusing on 'People'

Thanks to Claudie, Natalia reduced project management work that took 10-15 hours per week to just 1 hour. But this doesn't threaten her job. Instead, she provides feedback when AI makes mistakes and serves in a managerial role, building relationships.

Most importantly, the 'quality of time' has changed.

The part I love most about my work is being with people. I love our clients and enjoy spending time with them. Every hour not spent entering information into Excel sheets is an hour spent with the people I work with.

Natalia's case proves that AI, by automating simple repetitive tasks, enables us to focus more on inherently human values of 'communication' and 'creativity.'