Open letter to Sam Altman (respectfully, from the small business trenches)

Sam.

First, flowers. OpenAI is the original gangster of this whole LLM era. You lit the match, everyone else showed up with marshmallows, and now half the internet is selling “AI-powered” anything like it’s a personality trait.

And truly, we respect the pace, the pressure, and the responsibility of building the thing everyone is now building on top of.

But I need to say this out loud on behalf of the solopreneurs and small business owners who are trying to run real companies, serve real clients, and not spiral into a full identity crisis every time the interface sneezes.

Because at this point, ChatGPT feels like the most talented employee on earth… who keeps rearranging the office overnight, hiding the stapler for sport, and then acting surprised when everyone is late to the meeting.

We are not mad about innovation. We are mad about volatility. The constant updates and random UX changes are not “fun surprises.”

They are constant operational interruptions.

We build workflows around you. We train our teams around you. We finally stop micromanaging every tiny detail, like an Instagram caption, and then BOOM, something foundational changes and we feel like you put us in a chokehold.

You giveth and you taketh away, Sam.

One week, calling a GPT into an existing conversation works. The next week, it doesn’t. At one point, I could attach Google Docs straight from Drive like a functioning adult with a business. Then it vanished.

Not gone for everyone. Just for some. Which is a super fun game when your job is trying to build consistent systems instead of playing “Guess Which Feature You Get Today.”

And then there’s the current personal favorite: my GPTs getting wiped from my account like they were never there. I can still access them through direct links, like I’m sneaking into my own house through a window, but I cannot actually edit them.

That is not a quirky little glitch, Sam. That is a trust issue, and small business owners like me are running on approximately 3% trust capacity at all times.

Now let’s talk about the vibe shift. Model changes are real. Safety guardrails matter. But there has been some serious personality whiplash.

A while back, ChatGPT was helpful and sharp and had enough “human” to feel collaborative. Then suddenly it got argumentative, obsessed with pushback, and determined to take every prompt down a thousand rabbit holes like, “Have you considered redefining the meaning of the word ‘and’?”

Sir. I am trying to write a client email and maybe eat dinner before 9pm.

Here’s why this hurts so much for our world. Entrepreneurs and freelancers are already drowning in operational chaos, decision fatigue, and a thousand competing priorities.

Marketing alone eats half the week, and client work always wins. Delegation is hard because hiring is hard, onboarding is messy, and our last nerve is fragile.

Many of us are trying to stop making worse calls just because we are mentally done, and we are trying to build systems that make focus possible instead of living in “good opportunity” land forever.

So when the tool that’s supposed to reduce mental load becomes another thing we have to babysit, troubleshoot, and constantly re-figure out, it does NOT feel like progress, Sam.

It feels like more work disguised as “AI-powered productivity.”

And that’s the bigger point. Small business owners do not actually care about AI or productivity. They care about outcomes. They want reliability. They need solutions.

They want systems that help them stop being the kink in the hose, stop micromanaging, stop working until midnight, and stop treating their business like a never-ending group project with themselves!

(Would it kill ChatGPT to give us results that don’t create three new tasks every time it saves one?!!)

So, respectfully, here’s the ask: if ChatGPT is going to be the operating layer for modern work, it has to act like infrastructure, not a moving target.

Give people a stability mode. Give normal-human change logs. Be clear about feature rollouts instead of quietly making half the room feel crazy.

And for the love of all that is caffeinated, do not make custom GPTs feel like a hormonal teenager that only follows directions when they feel like and gives you the evil eye from under their hoodie when they don’t!

The setup is the easy part. Keeping it running is the whole job. And right now, too many of us are stuck “babysitting AI agents” when we are already stretched thin.

I want to stay loyal. A lot of us do. We are not here to burn it down. We are here asking for something wildly unreasonable in tech: a tool that works the same way today that it worked yesterday.

Respectfully,

A tired solopreneur who just wants to sip her coffee in peace, instead of accidentally knocking it over in outrage over yet another software update.

(true story…maybe…probably 🙄)