Deep Tech: Validate Business or Science/Tech Capability?

I've worked with real scientists for years as a technical programmer so have a humble appreciation of just how deep tech can go. Is this really innovative science?

I'd call Touchgram a moderately deep tech product, compared to a lot of startups especially on here. It lets you create interactive messages that respond to touch, including any gestures you specify in the message. The combination of game technology and gesture recognition is fairly unique (why I'm seeking a patent) and not easy (why I have grey hair).

I'm also a solo founder.

So I have a lot of sympathy.

But, I've also talked to a lot of founders in similar positions. If you haven't identified a starting market, it doesn't really matter how good your tech - it's still going to need a pathway to market. The only way to validate a tech capability (outside of scientific research) is to get it to market.

maybe I'm spending too much time on tech instead of business? almost certainly!

Much to my wife's despair, I've probably wasted over a year in the last five since I went through Founder Institute with Touchgram, on tech that's either now obsolete or won't be used in the product before 2021. (For the curious doing math, I took 2 years off to work at Realm, I've only been working fulltime since May 2017).

If you haven't taken a slice of your tech all the way through to a delivered product then I can guarantee you have overlooked a massive amount of boring, trivial work.

Usually, when we are talking about “deep tech,” we are referring to technology that requires basic scientific research before you even start building it. New hardware architecture, fundamentally new algorithm development, application of Nobel prize-winning ideas, etc. For this kind of product, lean/validation heavy methodology is hard to apply because every iteration could be years to create. If your deep learning project requires fundamentally new algorithm development than I would suggest creating the first working proto, publish It is a peer-reviewed journal to get validation and some buzz than find 1-2 client who needs it. My experience shows that 99% of deep learning application does not fall into this category since about 2015. If that is your case go to the potential client's first maybe do some consulting work for them and then create the product.

Your use of the terms is messed up, but if you worship Lean you'd ideally/generally validate starting from the problem, to market size, to reaching the market cost-efficiently, to willingness of the market to pay for your solution, to your solution actually solving the problem (and only after that start building the solution).