Jordan Sherer

This year's AI layoffs will turn into next year's hiring.

The loudest voices telling us to put our agents in a loop are also those selling tokens.

I took a break this weekend.

Determinism in AI doesn't come from prompts.

This week I added "Code Mode" to Factorly.

The "agent reads your .env" problem isn't an LLM problem It's a credential management problem that we've solved for decades in other contexts.

AI is going to drastically increase in price Once the frontier labs decide they need to make money...

Bug reports are at an all-time low...

We're currently in the "throw food at the wall and see what sticks" phase of software product innovation.

AI agents are not employees.

It's Wednesday and I've already used 125M tokens in Claude this week...

If you're not shipping, you're not learning.

Flexing how many tokens you've burned (yes, even lines of code generated) with AI is nonsensical.

"Please, please, please don't delete our production database..."

It's Friday.

Systems often fail because they are poor at representing accurate state And humans are poor at resolving ambiguous states under time pressure.

Your centralized judgment doesn't scale The effort to eliminate bottlenecks by being the "decider of everything" will only make you the bottleneck.

be kind. be bold. be brave. in that order.

You cannot just add more AI and call it a strategy.

Your tech is not your differentiation It never has been.

AI will flatten our orgs The middle.

Typing has never been the bottleneck.

"You don't look like you have epilepsy."

Bias in hiring isn’t always loud.

Company culture isn't branded swag and virtual happy hours.

We improve our critical thinking when we write.

We must stop celebrating hustle like it’s the path to success.

Stop asking “where do you see yourself in 5 years?”

Remember that the easiest way to validate an assumption about your market… … Is to talk to your customers about it.

We need to stop live coding in interviews.

Your AI strategy is only as good as your Data strategy.

AI will only take your job if you stop learning and adapting

You'll hear it from me repeatedly...

Human judgment is still essential for hiring great talent.

We’ll know we’re nearing AGI when I can use it to solve my Daughter’s “Where’s Waldo?”

Skilled humans directing the tool, not the other way around.

AI is a tool to eliminate the mundane so we can focus more on what makes us human.

I believe that complexity develops from simple components arranged to create unpredictable interactions.

Hiring should be easy.

DevOps is not a team… Frontend, backend, qa, data science are not teams.

Where is your team going?

Stability is an illusion… Optimizing for perfect conditions creates fragile systems and those cause cascading failures.

You should never use microservices...

Servant leadership belongs in startups.

Every startup scrambling because OpenAI has been down for more than a few hours is not an “AI company”

"Septillion years of classical computing"...

You can write more effectively in 15 seconds… Write the most important thing.

A candidate’s work history doesn’t matter… No, it’s how they adapted during that history, what they failed at and learned from, and how it all has shaped them into who they are today.

It’s the weekend.

There’s nothing worse in a presentation or sales call or meeting or email or slack message than using meaningless adjectives to fill the air.