Will you take a quick look at this thing I’m building? I get this question on a daily basis.
Over the last 12-18 months, the goal posts on the level of effort to build a passable side project has dramatically decreased. It will certainly continue to decrease. Ironically by lowering the bar to actually ship something, the frontier models have simultaneously raised the bar on how these things are evaluated.
As this bar has lowered there are more individuals asking themselves if their fun side project could be the next big thing.
It used to be if you just shipped something it went a long way in terms of a positive signal that you were a reasonably capable person. Shipping a working product was no trivial thing. In terms of a success bar you were something like a minimum 80th percentile if you just got the thing live. Beyond technical literacy it demonstrated basic signs of grit and determination that are required to be a successful entrepreneur. If your side project actually made it to revenue prior to funding, it was relatively easy to raise capital.1
The goal posts have shifted significantly.
Now the baseline expectation2 has become that your shipped product has serious signs of traction with users and likely some initial revenue. In a world where everyone is inundated with the steep slopes of AI company time to revenue3, the expectation is that this should happen with all startups.
Enter mid-2026: A half baked AI built product with no traction likely raises so many doubts it would almost be better if you didn’t have a product4.
Here is my quick mental checklist5 I run through when I’m sent a project looking for investment or inclusion into our venture studio:
Product sense: is this person able to identify real pain points? does this product solve a real problem?
Grit: how much effort was required to ship this? did this person create something basic after two prompts and then stop? How long have they been working on this?
Technical proficiency: is the technical approach so obviously wrong for the product it would stunt adoption?
Traction: is anyone using this?
Domain expertise: if this is the right person to deeply understand this problem, why weren’t they able to immediately get 10 people on the phone the next day to take a look?
It has become so much harder to tell good from bad, trying to dig into the product and use it is the only option and it takes more time than ever. Really though, the evaluation is the same as it has ever been: if I extrapolate out this level of effort at this rate of progress, what’s going to happen?
So how can you avoid some of the more common pitfalls in an age where everyone can stand up an mvp in an afternoon? Besides the obvious non-helpful advice (just make something people want! have obvious signs of traction), here are a few concrete suggestions to consider about the project you are pitching:
Avoid a giant scope vs vision mismatch.
If you have a tiny narrowly focused wedge product to validate a problem space, that’s great.
If you have grandiose marketing language around a bunch of features that don’t exist, that’s bad.
Show consistent signs of effort and progress.
This could be as simple as patch notes or a changelog.
It could be by posting #buildinpublic content6.
Or it could be in your follow up where you explain to me “hey since we last connected, I spoke with these users and shipped xyz features.”
Clean up the obvious hallmarks of AI driven design.
Pick an aesthetic that isn’t blue/purple on black.
Remove all the superfluous copy.
Make your marketing site be more than a single page.
Make the product actually work.
Saying you created a product when you actually only built a marketing page is beyond lazy.
At this point I just assume a waitlist or form to learn more means nothing actually exists in terms of working product.
The product must be well designed
Generative AI has become so good in this area, poor design is now inexcusable
This includes obviously broken flows, orphaned pages and incomprehensible configurations
Frame this project as one of many and show a long list of things you’ve built.
This is probably the one thing you can do to lower the bar you’re evaluated on.
On the final point, framing this as one of many projects or the one you are most excited about is a good way to set appropriate expectations that you aren’t presenting the next Anthropic. You are a curious person trying different things and you understand what it takes to try different ideas to achieve product market fit.
I’m biased here but I believe having a long history of interesting projects you’ve executed on is a valuable datapoint to prove yourself as an individual. At the end of the day, investors are evaluating the person first and the product second. Super early products will likely change and evolve. Early stage investments are on the team and if they can navigate those changes successfully.
To that end, I also try to walk the walk in relation to advice I offer so here’s some highlights of projects I’ve been working on in my free time.
In aggregate, this is what the growth looks like month over month for all 18 projects I’ve shipped in the last year or so:
Some of the more recent and more popular projects:
RaceTrackDesigner - a fun tool to design race tracks on any real road in the world
Lesmo - the best place to find and rent a race car in Europe
Llama Farmer - an intentionally borderline infuriating management sim where you run a llama farm
Stylometry Concealer - an experiment around anonymity in the age of AI, inspired by the most recent Satoshi accusations7
Gmail Janitor - open source tools to automatically clean out old gmail inboxes that are hitting storage limits
All of these are in the context of an ever growing list on my quirky personal web site.
But still not that easy.
Overgeneralizing, team still matters a ton.
So much of this has unfortunately now been gamified and that activity is further skewing reasonable expectations.
But if you don’t have a product you are too early and not committed enough because it’s so easy to build now. Investors like to have their cake and eat it too.
This ignores things like team which matters
If you make this content and it’s good, bless you. My attempts have been so painful and cringe. I’m not cut out to be an influencer.
Article in question - for what it’s worth, I’m still skeptical but I find the use of AI to match old anonymously written bodies of work to be interesting
Did you write this for me? Great article Kyle! Always enjoy your writings. Will recommend to my circle.
Good article. I need the gmail janitor.