One Woman Show:
AI Edition
I have a team. They're not on payroll. They don't sleep. And once a week, they report to me.
I have a team. They're not on payroll. They don't sleep. And once a week, they report to me.
Most designers go solo. Every project, every brief, every revision, one person doing it all. The one woman show. I've been that person. I still am.
But somewhere in the last few months, the show got a crew. Not collaborators. Not freelancers. Agents. Six of them, running in parallel, each with a specific role. Some run every day on a schedule. Some I wake up when I need them. Once a week, I sit down and they all report in.
This is what that looks like from the inside.
Runs every morning at 06:01. Scans Reddit, HackerNews, and LinkedIn for relevant signals: AI design tools, portfolio trends, competitor content. Scores each item, saves it to a SQLite database, sends me a morning digest. This week it got smarter: LinkedIn scores moved from a hardcoded 7.5 to real content-based scoring, with a bonus for curated picks.
21 items in DB this weekManages the entire portfolio site. Knows every file, every deploy step, the bilingual structure (EN + HE), the design system. Has a living document (CLAUDE.md) it reads at the start of every session so it never forgets where we left off. Nothing goes live without my sign-off.
26+ files under active managementTracks the Portfolio Manager landing page: signups, copy experiments, messaging shifts. Opens each session with a status brief. Part product manager, part editor.
Waitlist + copy optimizationKnows my positioning, my audience filters, my category language. When I'm writing a LinkedIn post or deciding on a content angle, this agent holds the strategic frame so I don't drift. It's read STRATEGY.md. It remembers what I decided last week.
Strategy + content direction14 versions. That's how many cuts of the Portfolio Manager teaser video we've made together. Artidansel handles timing, transitions, text animation, and iteration. v14 is the current candidate. We're not done.
v14 current / iteratingA scheduling engine built for a school. It runs independently from everything else: its own codebase, its own logic. The agent handles updates, edge cases, and fixes without context-switching me into another project every time something changes.
Autonomous / client-facingIt started as an experiment. I opened a session and asked: "Brief me. What happened this week?" Each agent had something to say. The Hunter shared its scoring update. The Site Manager listed what changed on the site. Artidansel flagged that v14 was ready for review.
It didn't feel like talking to software. It felt like a standup.
I have a coffee. They have the data. The briefing takes about 20 minutes. Then I decide what moves next.
I opened a session and asked the team for their weekly report. They reported. I looked at what they'd produced and thought: this is a case study. The workflow, the agents, the briefing itself. It's all already here.
I didn't write this from scratch. I extracted it. That's exactly what Portfolio Manager is designed to do: the case study lives inside the work. You just need something to pull it out.
"I'm not using AI. I'm managing an AI team. There's a difference. One is a tool. The other is a practice."
Merav Shaked
The decisions are still mine. The creative direction, the client relationships, the taste. What the agents handle: the daily scanning, the repetitive updates, the iteration cycles, the stuff that used to take hours and now takes a briefing.
One woman show. AI Edition.
Tell me what you're working on. I'll be in touch.
You'll hear from me when things move forward.