The 2x-40x AI Enhanced Consultant

Three and a half years ago, ChatGPT came out and doubled my productivity. I have come a long way - AI has progressed even more. I started by talking with ChatGPT, asking it to write code, copy the code out and put it in VS Code. Run, find bugs, copy those back into ChatGPT. That was all anybody could do at first. That already got me to at least double the amount of coding work I could do. That is nice at 2x, but today it feels more like 40x.

Now I am solidly in the “running/managing/directing a team of agents” phase. It’s like having an entire staff of customized specialists able to do work for me. It’s so much like managing humans that I’ve moved to calling them AI employees instead of agents. I am working on a dozen projects. Typically 4 or so at any given moment. I’ll set up a mission or a goal and set one of my AI employees to over see a set of other AI employees to accomplish the task.

That’s right, I’m managing the managers.

I’m not talking about a dozen to do items. What I mean is this…

On any given day, I have a dozen independent initiatives moving forward simultaneously. Not ideas. Not TODO items. Actual software being designed, coded, tested, documented, reviewed, and in many cases running autonomously.

A snapshot from one week looked something like this:

• An autonomous RFP Factory running hourly on Linux.
• A commercial-quality data modeling tool that supports forward and reverse engineering into Snowflake and has a built in assistant where you can ask/direct new tables and relationships and it creates them.
• A Snowflake RBAC governance analyzer producing executive consulting reports.
• A Career Advisor application interviewing users and generating resumes that will help someone I know land work.
• The AI Employee Factory itself evolving through autonomous ideation.
• Agent-Orch and Auto-Orch continuously improving their orchestration capabilities.
• A Chief of Staff AI employee managing knowledge and tracking projects.
• Documentation, marketing videos, websites, books, consulting methodologies, and customer demonstrations all advancing in parallel.

None of those projects are waiting while I work on another one. They’re all moving. I get one going - it may take 30 minutes to an hour. Sometimes the AI employee will work for a few hours. Then I move and start the next. More than 4 efforts running simultaneously and I start losing the ability to keep them straight. But over the course of a week, I’m easily moving forward on a dozen projects. And I’ve trained the AI employees to keep track of what they are doing, the mission and what’s next. So even though I struggle to keep track, I can have a status meeting and be brought back up to speed.

People assume AI is replacing programmers. It isn’t replacing me. It has changed what my job is. Ten years ago I spent my day typing. Today I spend my day thinking, Designing, Reviewing, Redirecting, Approving, Rejecting, Inventing.

When one AI employee is implementing a feature, I’m architecting the next system. While another is documenting yesterday’s work, I’m sketching an entirely new product. When a long-running coding session starts, I don’t sit there watching tokens scroll across the screen. I switch to another project.

Or I SSH into my Linux box from the couch with my iPad, reconnect to the persistent terminal session, review the completed work, give new direction, and send the employee back to work.

Then I go back to watching television while my AI employees keep working.

That’s a very different way of building software.

AI employees don’t all have the same job

This is another misconception. People imagine one AI assistant. I don’t. I have specialists.

One is reviewing architecture. One is writing code. One is creating documentation. One is producing marketing material. One is researching. One is reviewing another AI’s work. One is managing projects. One is helping me look for customers. One is building the tools that build the next AI employees.

Just like a consulting company doesn’t hire twenty project managers. It hires specialists.

The multiplier isn’t typing faster Typing faster might be worth 20%. Maybe 50%. That’s not where the productivity comes from.

The multiplier comes from parallelism. It’s not me using ChatGPT, it’s me managing a team of managers who are directing specialists.

The AI Employee Factory is dogfooding itself One thing I love is that the platform is building itself. Agent-Orch improves Auto-Orch. Auto-Orch builds new applications.

Those applications improve the development process. The Chief of Staff tracks all of it.

The knowledge base gets better. The orchestration gets better. The reports get better.

The next project starts from a stronger foundation than the last. It’s a compounding effect.

I’m between jobs. I lost my job as Director of Data and AI because I wasn’t a project manager. I’m looking for work, but I’m also moving head long into just running my own AI empowered consulting company. What I’ve put together in a month is ASTOUNDING. I’m no mere project manager, I’m a solution’s architect who’s super powered by a company of AI employees.