Memory systems for products like Claude Code and Codex might not be as reliable as they seem because they're heavily dependent on the harness, how the LLM is guided to retrieve and use them.

More customizable agents like OpenClaw and Hermes solve some of this with plugins and custom memory systems.

But then you need memory that works across devices, local sessions and cloud agents.

So I'm wondering : is AI memory actually a product, or just infrastructure agents will eventually handle themselves?

I've been thinking a lot about what happens when AI agents move from short tasks to workflows that last days or even months.

Longer context windows help, but I'm not convinced they solve the whole problem.

For example, imagine an agent working with the same customer for six months.

It might need to remember that:

Something I've never been able to work out. When someone tries a generative tool for the first time, does the licensing question come up before they generate anything or only once they've got an output they actually want to use?

I ask because I had to decide where to put that line, and every option had a cost.

The export screen felt safest. But that's the one moment where the person is happy with what they're looking at, and dropping legal text into it turns a good moment into a form. Rejected.

Terms page only is what most tools do. Nobody reads it, and "it's in the terms" is not an answer to someone deciding whether to spend an afternoon on you. Rejected.

I've been exploring better ways to validate early-stage product ideas not just from gut feel, but through meaningful market signals.

In your experience:

• How do you go about validating a new product idea or GTM direction?

• Do you lean on user interviews, market data, experiments, or something else?

• What s worked well (or not so well) for you?

As a product person myself, I ve always found this step to be the trickiest especially when time and budgets are tight.

If you take money in a zero-decimal currency through Stripe, there is a way to lose a payment that looks like nothing is wrong. I lost weeks to it, so here it is written down.

Stripe has sixteen currencies with no minor unit, yen among them. The amount you send is the amount, not hundredths of it. That part is documented and everybody reads it.

Obviously spending money on advertising is mostly out of the question (at least for me), so I'm struggling to find ways to show real potential users my product.

Except for all the standard stuff (generic subreddits, Discord servers, sites like this one, nice Github page, etc), what has been successful for you?

This evening we have our Demo Night in San Francisco, and our team is stoked to see some amazing products in action later on! If you're in SF - come out and say hi!
Here's the event: https://lu.ma/demos

Bought this domain long back, thinking between selling or using this for domain for all my open source projects.

Just wanted to know what is it valued in the market.