A talk by Viktor Grant

Describe It. Read It. Fix It.

I can write the code myself.

Now I describe it, and the AI does the heavy lifting.

Invite me to give this talk

The picture

See It First

Tap a point to jump to it.

The idea

What I Say

I can still write it, design it and author it.
Nearly all of the heavy lifting is done by the AI.

My part is simple to say and hard to do.
I describe what the code is supposed to do.

Then I read what came back, and I fix it.
Reading code is the skill that matters now.

This is how a whole site gets rebuilt in days.
One person, one clear description at a time.

What we built lately Tools made this way, this month.

In the room

What People Leave With

How to describe behavior so the AI builds the right thing.
How to read code you did not write, without fear.
A loop you can run on your own project tomorrow.

For a panel

Questions I Would Bring

What can you describe clearly that you cannot build yourself?
Who checks the work when the machine writes it?
What did you learn from reading code you did not write?

A good question beats a prepared speech.

Your stage

How This Fits Your Program

First choice

Panel

I show what I asked for, and what I fixed.

Hands on

Mastermind room

Everyone describes a small tool. We read what comes back.

On request

Keynote

A live build: describe, read, fix, ship.

Invite me

Put It On Your Program

Tell me the date, the room, and the question you want asked.

Invite me to give this talk All talks About Viktor

Keep going

Other Talks

Your People, Your Data Reach your audience without renting it back. Feed Your AI Five Times What You Ask The one habit that makes AI answer better. The Gap Between The Claim And The Real Thing Every system has two versions of itself. Your AI Forgot Yesterday Write the memory yourself. Hand it over as files. AI For Authors, In Your Own Voice Use AI without losing the person readers came for. The Room Is On Zoom Lead a group online so people come back. Every Username Is A Node Build a network that outlives the builder. What A 6502 Taught Me About AI Tiny memory made me careful. AI context is memory too. The Future Of Work, From Someone Who Never Had A Job Create real value, and find a way to measure it. Your AI Is Not An Employee It is a pattern synthesizer. Use it like one. Stop Chatting. Start Building. Make the wrong answers unavailable. How I Work With Claude One person, one live site, and a loop that checks the work. One Workbench For Every AI Prepare the work in one place. Carry it to the AI that fits. The Private Island Build in private. Show in public. Think Cheap. Publish Polished. Two kinds of model, and why using both saves tokens. A Year In 1994. A Week In 2026. We invent the future, then wait for the tools to catch up. One Address For Every Relationship Relationships are Einstein. E-lationships are Bucky Fuller. If You Cannot Reach Your Members, You Have No Community A community is the people you can reach.

Which one stuck?

Tell me which talk you would put on a stage.

Message me All talks