What Happened to ClearSteps AI

On Experience

Aug 20

For a while, I was trying to build a business called ClearSteps AI.

There was an LLC. There were expenses. There was a bank account and paperwork and all the other little administrative ceremonies by which an idea becomes, at least technically, a company.

There was no revenue.

That fact is funny to me now, although probably less funny to the version of me filling out the tax forms.

But, ClearSteps was real. Just maybe not in the way a business is supposed to be real.

The basic idea was simple: AI was becoming enormously powerful very quickly, and most of the conversation around it seemed to assume that everyone either understood what was happening or was willing to spend several hours figuring it out.

I wanted to help the people who weren't.

Not engineers. Not necessarily people following every model release or arguing about benchmarks online. Ordinary people. Parents. Older adults. People with disabilities. Small-business owners. Someone who had heard that ChatGPT could be useful but had no idea what they were actually supposed to type into the box.

ClearSteps was supposed to make AI less intimidating.

One clear step at a time.

The Idea

I had several versions of what ClearSteps might eventually become.

There was tutoring and consulting. There were articles. There were videos and lessons. At one point I considered publishing free material regularly and putting more substantial material behind a paywall. I thought about helping organizations use AI responsibly. I thought about teaching people how to use these systems not as novelty machines but as practical tools.

Accessibility was always part of it.

That wasn't an abstract concern for me. I'm legally blind. I have spent years learning which technologies actually make something easier and which technically qualify as "accessible" while still making you want to throw the computer through a window.

So, when I thought about AI education, I thought about voice input. Read-aloud tools. Larger interfaces. Fatigue. How many steps something required. Whether somebody with limited technical experience could actually follow along.

And, I thought a lot about people like my own family.

At various points, ClearSteps became less about explaining "artificial intelligence" and more about showing someone what they could actually do with it.

Find information.

Compare options.

Ask follow-up questions.

Take an answer and add the context that changes it.

Use AI as something closer to a thinking partner than a strange search engine with a personality.

That part still seems important to me.

Maybe more important now.

The Problem With Building a Business Around Help

There was always a strange tension inside ClearSteps.

For it to succeed as a business, people had to continue needing help using AI.

But, I wanted AI to become easier for everyone to use.

I remember worrying, especially as the systems improved, that they might eventually become simple enough that ClearSteps would lose its reason to exist. Models were getting better at understanding ordinary language. Interfaces were improving. Things that once required elaborate prompting were becoming increasingly natural.

And, I caught myself realizing how absurd the worry was.

If AI became so intuitive that an elderly person, someone with a disability, or someone who knew absolutely nothing about technology could simply talk to it and get useful help, that would be wonderful.

It might be terrible for my business.

But, it would be wonderful.

There is probably a lesson somewhere in there about building a company around a problem you sincerely hope disappears.

I never resolved that tension.

I don't think I needed to.

Something else resolved it for me.

When I Couldn't Keep Talking

My health had already been changing what was possible.

Family mattered more. Energy mattered more. ClearSteps gradually became harder to imagine as something I could continually feed with the amount of work a real business requires.

Then speaking became harder.

Eventually, it became hard enough that I couldn't keep recording.

There wasn't a grand strategic meeting where I decided ClearSteps AI had reached the end of its useful life. I didn't sit down with a spreadsheet and determine that the market opportunity wasn't large enough.

I just couldn't do the thing anymore.

That distinction matters to me.

I was forced to stop.

There is grief in that. Speaking itself had already become something I was losing, and ClearSteps happened to depend on the exact ability that was becoming unreliable. Recording lessons requires talking. Talking requires a degree of control that I no longer had.

So, the decision was made for me.

And, now that enough time has passed, I can admit something slightly uncomfortable.

Maybe that was good.

Not the difficulty speaking. I would take fluent speech back without hesitation.

But, the stopping.

Because when I imagine what continuing ClearSteps as a genuine business would have required, it sounds almost nightmarish.

What Actually Failed?

Calling this a postmortem makes ClearSteps sound dead.

I'm not sure it is.

The business is.

Or close enough that I don't feel any need to perform the distinction.

ClearSteps AI never became a meaningful source of income. I spent money on it. I spent time on it. I incorporated it. I made plans. I tried things. Then I stopped.

By conventional business standards, this is not a difficult case to classify.

It failed.

I'm surprisingly comfortable writing that.

Because the more interesting question is what, exactly, failed.

The LLC?

Sure.

The business model?

Definitely.

The idea that I should turn this particular interest into another obligation that needed to justify itself financially?

Probably for the best.

But, the reason I started it is still here.

I still care about whether ordinary people can understand what these systems are doing.

I still care about accessibility.

I still find myself helping family members figure out how to use AI for something practical and then explaining why one approach worked better than another.

I still care much more about what this technology means for somebody's actual life than whether a new benchmark moved three points.

And, I am still writing about AI.

Just differently.

What Survived

Some of ClearSteps ended up here, in Open Doors.

Not as a business.

Not as a funnel toward anything.

Just as one of the things I think about.

I even briefly considered letting Open Doors end. Christa stopped me. She was right.

That difference has turned out to matter enormously.

I can write about AI when I actually have something to say about it. I can write about a model release and then write about writers, disability, my children, astronomy, consciousness, or some strange thought that occurred to me while sitting in the kitchen.

Nobody has to convert.

There is no customer journey.

Thank God.

And, when someone in my family wants to understand how to use one of these systems, I can still help them. There doesn't need to be a product attached to the interaction. The useful part of ClearSteps was never the LLC.

It was the impulse behind it:

Here is this complicated thing.

Come here.

I'll show you.

That survived completely.

The Part I Would Keep

I don't regret ClearSteps.

I don't particularly wish I had tried harder to turn it into a successful company either.

That's a strange thing to discover about something you once took seriously enough to legally incorporate.

But, maybe ClearSteps taught me that not every meaningful interest needs to become a business.

There is a peculiar pressure in American life to monetize whatever you are good at or care about. If you write, sell the writing. If you teach people something, build a course. If people find your explanations useful, create a subscription tier.

Maybe.

Or, maybe sometimes you can just help someone.

ClearSteps began because I saw a gap between what AI could do and what many people understood it could do. I wanted to make that gap smaller, especially for people technology too often leaves behind.

I still want that.

The company wasn't necessary for the wanting.

In fact, losing the company may have made the original idea clearer.

I couldn't keep recording. Eventually, I couldn't keep building ClearSteps in the form I had imagined. My body closed that particular door before I was ready to close it myself.

But, what mattered walked through somewhere else.

I still explain.

I still experiment.

I still help my family.

I still write.

ClearSteps AI was supposed to teach people that these strange new machines could be useful in ordinary human lives.

I don't need a business to do that.

Turns out I never did.

It was meant to help
Something official to help
Did it die? Not yet

Kevin Gleich