$ ./openclaw present --venue era --date 2026-03-24

Make AI Work For You

or: How I Stopped Prompting and Started Delegating

Chuck Blake // ERA // March 2026

You Are Either Drowning Or Ignoring It

There is a third option

AI is moving at an inhuman pace
Most people: infinite tabs, no workflow
Others: "I tried ChatGPT once"
Neither is working

[ IMAGE: drowning-or-ignoring ]

Split screen: chaotic browser tabs on left, person ignoring laptop on right. Hand-drawn / editorial style.

The goal is not to use more AI. The goal is to get more done.

A Personal AI Chief of Staff

Not a chatbot. A chatbot answers questions. A chief of staff gets things done.

Runs on my machine — my data, my rules
Connected to Telegram, my calendar, my code, my tasks
Always on — memory, tools, real actions
Has his own name (Gomez), his own vault, his own rules

[ IMAGE: gomez-diagram ]

Simple diagram: Chuck → Telegram → Gomez → Tools (GitHub, OmniFocus, Email, Heroku)

It Did Not Go Perfectly

Shocking, I know

Built too many systems at once — none of them solid
It confidently marked things "done" that were not done
A sub-agent recreated a cron job I had just killed
Morning briefing fired at 5pm

R14 — Memory quota exceeded (287 accidentally re-queued jobs)

The recurring theme: I built something, it seemed to work, I moved on. It was not working.

Thing 1: What You Feed It Matters More Than The Model

Garbage in, confident garbage out

Raw prompts get generic output
Curated context (your vault, your errors, your preferences) gets your output
Build a personal knowledge base — notes, highlights, decisions
The model is a commodity. Your context is the moat.

[ IMAGE: inputs-diagram ]

Funnel: Readwise highlights + personal notes + errors.md → AI → personalized output

Thing 2: Build Once, Use Forever

The difference between a tool and a system

A skill is a reusable instruction set for a specific task
good-morning, deploy, process-inbox, grill-me...
Not model-specific — works across Claude, GPT, whatever is next
~75 skills in my setup. Each one took 20 min to build. Each one saves hours.

Agent orchestrates — scripts execute.

Thing 3: Your Thinking, Made Portable

If it is not written down, it does not exist

Four layers: identity, errors, long-term memory, daily log
Semantic search via local vector DB (Mem0 + Qdrant)
Key insight: this is YOUR thinking, not the model's
When the next AI platform comes — you bring your memory with you

The agent wakes up fresh every session. The files are the memory.

Thing 4: The Difference Between Vibes And Knowing

You cannot improve what you cannot measure

An eval is a test for AI output
Without evals: improving by feel ("seems better?")
With evals: you have a score. You can iterate, regress, ship with confidence.
Autoresearch: runs a skill 50x, scores output, mutates the prompt, keeps improvements

You cannot improve what you cannot measure. This is true for software. It is especially true for AI.

Thing 5: Are You Running The Factory Or Working On The Line?

Warhol did not apologize for running a studio

FORD'S FACTORY

Model T: 12 hrs → 93 min.

Interchangeable parts.

Human as plant manager.

WARHOL'S FACTORY

Vision from Warhol.

Assistants executed.

He made the production system the aesthetic.

The bottleneck used to be execution. Now it is judgment. That is the right problem to have.

Make AI Work For You

One thing you can do this week

Pick ONE repetitive task you do every week
Write down exactly how you do it (step by step)
That is your first skill
The system grows from there

AI is moving fast. Most people are drowning or ignoring it. You now have a third option.

chuckblake.com/presentations/openclaw-era-v5/

Find Me

I am probably already on your machine

Website: chuckblake.com
X: @chuckblake
LinkedIn: /in/chuckblake
Slides: chuckblake.com/presentations/openclaw-era-v5/