[k3nt0456]

Feels like we're at the stage where models are good enough, the missing piece is tooling which still needs to catch up to make full use of them.

For coding, this has started to happen and it's in full swing, what about other domains?

[teruakohatu]

> For coding, this has started to happen and it's in full swing, what about other domains?

There needs to be a moat. I see all these startups with a moat that is no bigger obstacle than a puddle.

A local (geographically to myself) pre-LLM (Convolutional Neural Network) company invested in a huge moat, building a large training set of human labeled data ... I knew the moment I saw an LLM that could ingest image that their moat had evaporated in an instant. A lot of the technical staff either left or lost their jobs but the company appears to be limping on hoping for an exit.

If a startup depends on a secret prompt and a few MCP servers, they have no moat.

I think the next wave of AI-based SaaS (rather than pure AI API providers) will be companies like the Legal IDE (Tritium [1]), recently featured on the homepage. Tools that innovate and use AI, but would still be innovate even without it.

[1] I have no affiliation whatsoever. I just liked the demo and the concept.

[neom]

I was talking to Jeff Lawson the other day and he said something really interesting I hadn't really considered the nuance. I always considered Twilio, DigitalOcean, Sendgrid etc "dev tools" - but Jeff said I was wrong and they are supply chain primitives/components. DevTools are more like IDEs in this lens.

Only reason I bring it up is related to your moat comment. supply chain stuff is tied into workflow that becomes institutionalized, and typically more b2b, dev tools are more like b2c, so you really have to compete on merits hard, moats of b2c2b look to Slack or anything with seats/licenses/team features.

In the new AI world we can see it starting to emerge, on one hand you have something like Cursor, on the other something like charlielabs.ai - Will be interesting to see how it all plays out, but I wanted to just add some nuance I've been thinking about recently. :)

(full disclosure I'm considering joining charlie labs, don't want to be accused of shilling later if I do.)

[forrestthewoods]

> There needs to be a moat. I see all these startups with a moat that is no bigger obstacle than a puddle.

Cursor doesn’t have a moat. And yet…

[tough]

Cursor's moat is it's user's data.

they have SOTA completion models (bought NineTab, founder left already, but still).

Cursor moat is they're the touch point with customers, not the big labs. imo

they can use all that data to better improve the ux/ui and piggyback on general models improvements

[dangus]

Have they done that? In my opinion they’ve built an AI IDE consisting of 90% forked open source code that has an identical experience to competitors with less of a chance to integrate with big tech Google/Microsoft/Apple-style ecosystem products.

The “step one: get user data, step two: ???, step 3: profit” business model is well established, and it definitely doesn’t always work.

[flyinglizard]

The guys right in front of the user usually win. The rest are just price-sensitive commodity.

[dangus]

The difference is that this time the price sensitive commodity is right in front of the customer with very little value surrounding it.

[BoiledCabbage]

> There needs to be a moat.

Why does there need to be a moat? You don't believe in markets and competition?