About Amanda
The portfolio is the credential.
Written credentials are the signal AI just commoditized. Verifiable built work is the signal that gained value.
What I do now
I run a production multi-agent AI system every day, & I deploy the same methodology inside client teams.
Eleven agents with defined roles & real permission boundaries. Overnight verification agents check the system's own work before I read anything in the morning. It runs my business: my planning, my content pipeline, my client files. When I tell a client how to sequence an AI rollout, I'm describing something I maintain, not something I read about.
The evidence
The biggest one I've run: an enterprise knowledge-management implementation at Mural Consulting, April 2021 to May 2022. 150+ employees, 8 departments, 99% adoption, $0 budget.
The zero is the part worth reading. It's the constraint most teams are actually under, & it's why the method carried the result instead of the spending. The platform was Notion on the free tier, one paid admin account. The adoption number came out of a phased rollout & two years of weekly training sessions I delivered myself. When it hit resistance in month one, I went & learned change management & rebuilt the approach around what people told me.
The tools were different. The adoption problem was identical to the one AI rollouts have now.
What I've shipped
Content Loop, a production SaaS content system for podcasters, went concept to production in 10 days. Stripe checkout, trial lifecycle, recovery sequences.
The custom MCP server is a live sports odds API wired into Claude Cowork: TypeScript, six tools, built in an afternoon from the-odds-api.com V4 documentation. Public on GitHub.
Life Coach OS is the 11-agent system described above. Selected for the Agent Madness competition, 1 of 64 entries, advanced to Round 3.
The Implementation Lab Notebook is 26 entries & counting, public & dated, including the failures. Entry 002 is AI fabricating a work story that sounds exactly right. Read it.
I hold 4 Anthropic certifications, all proctored & identity-verified: AI Fluency, Claude 101, Cowork, & Claude Code in Action.
The path
I started in tech support. I spent four years running operations for a Microsoft partner team, where the job was the distance between the tool existing & the team using it. Now I run an independent AI implementation practice.
The problem hasn't changed. The tools did.
How I work, stated plainly.
I use AI to draft. This page, the newsletter, & the notebook all start as AI-assisted drafts & then get rewritten by me, which is the same workflow I teach clients. Where a notebook entry is substantially AI-drafted, the entry says so.
I have no commercial relationship with any AI vendor. No affiliate links, no referral fees. Nobody pays me to recommend a product, & I don't resell one.
Every number on this site comes from work I did. The Mural figures come from that project's closeout brief & its adoption reporting. The Content Loop & agent-system numbers come from my own run logs. The certifications are verifiable with Anthropic. If you want to know how any of them were measured, ask me & I'll walk you through it.