4 ms·

The best products happen at the intersection of an existing niche you are some level of expert in already and a technology that has not yet been adopted in that

The best products happen at the intersection of an existing niche you are some level of expert in already and a technology that has not yet been adopted in that niche. I've found one approach work very well with my mentees: - Figure out which "special interest groups" you are part of beyond software engineering. That can be "aquarium owner", "coffee lover", "morning person", "diligent grandson" — the less technical, the better. - Among these "niches", find the ones that could benefit from a transfer of technology, like (spitballing here) "teachers who work from home" (education niche) + "automated submission and pre-grading of homework" could work (digital document collection and rule-based checking logic), or "aquarium owner" + "nitrate cycle tracking IoT device" (hardware-enabled analytics) + "optimal light scheduling" (machine-learning-supported recommendation engine). Do that for all the groups you're part of, and you will find lots of ideas that aren't just "scratch-your-own-itch". They are 'scratch-an-itch-you-understand-and-know-how-to-remedy'.

christiansakai 6y ago
Aquarium owner here. Thinking of IoT hardware but turns out the sensor is expensive and error/breaking prone. Thinking to create yet another social media for people to post their aquariums and track the maintenance schedule then I realized I won't have the bandwidth (time and money) to deal with storing massive amounts of pictures, or policing the content to make sure it doesn't go down like social media dumpster fire of porn, gore, politics, religion, and other offensive materials
- MattGaiser 6y ago
  Take the Google Chrome approach and just have the AI randomly ban posts at a high rate.
  - graham_paul 6y ago
    You don't need "AI" for that
    - MattGaiser 6y ago
      If you call it that you can raise a million dollars first.
  - christiansakai 6y ago
    LMAO!
imhoguy 6y ago
Is it only me here who experienced such paradox: I love tech side-projects but often I don't want to introduce more tech in my non-tech hobbies and activites as they are remedy for too much screen time. E.g. my gardening hobby highest tech is $25 weekly schedule watering valve and I feel very good about it.
DoreenMichele 6y ago
"teachers who work from home" (education niche) + "automated submission and pre-grading of homework" I homeschooled for a lot of years and my first blog was a homeschooling blog. I was involved with The TAG Project as part of supporting my homeschooling and I really hate this idea. "Teaching to the test" is terrible teaching. Multi-guess answers aren't good ways to test student knowledge. They are just convenient for a teacher who needs to grade 25+ students. Please don't run around thinking of yet more ways to making it nominally more convenient and easy for individual teachers to process tests or homework or whatever from large numbers of students. This is not a way to enhance the transfer of knowledge to the future generation and is also not a good credentialing method. There are cool things being done with tech in the education space. There are people going online and learning things they want to learn because there are rich materials available for free. But they don't look anything like "pre-grading of homework" for teachers.
- lgregg 6y ago
  Yeah, but you're making an assumption, who said it had to be multi-choice answers? Why can't someone build a NLP model to test essays of middle schoolers or high schoolers graders on American history? You can look for learning objectives in the essay and grade upon that. Why not build a tool with D3 or another modeling tool to teach chemistry? You could have them build compounds and then if they make a mistake, insert via a tool tip and tell them why. As they combine things, you read out what it is or test them based on the string they put in for the compound. You have to build it and name it correctly.
  - DoreenMichele 6y ago
    Why not build a tool with D3 or another modeling tool to teach chemistry? You could have them build compounds and then if they make a mistake, insert via a tool tip and tell them why. As they combine things, you read out what it is or test them based on the string they put in for the compound. You have to build it and name it correctly. This sounds really cool and is completely unrelated to what I am saying "Please don't do."
- arvidkahl 6y ago
Diesel555 6y ago
I guess that's what I did. Niche is fighter pilot, product is generating target imagery for training. dieselplanning.com.

christiansakai 6y ago

Aquarium owner here. Thinking of IoT hardware but turns out the sensor is expensive and error/breaking prone. Thinking to create yet another social media for people to post their aquariums and track the maintenance schedule then I realized I won't have the bandwidth (time and money) to deal with storing massive amounts of pictures, or policing the content to make sure it doesn't go down like social media dumpster fire of porn, gore, politics, religion, and other offensive materials

MattGaiser 6y ago
Take the Google Chrome approach and just have the AI randomly ban posts at a high rate.
- graham_paul 6y ago
  You don't need "AI" for that
  - MattGaiser 6y ago
    If you call it that you can raise a million dollars first.
- christiansakai 6y ago
  LMAO!

imhoguy 6y ago

Is it only me here who experienced such paradox: I love tech side-projects but often I don't want to introduce more tech in my non-tech hobbies and activites as they are remedy for too much screen time. E.g. my gardening hobby highest tech is $25 weekly schedule watering valve and I feel very good about it.

DoreenMichele 6y ago

"teachers who work from home" (education niche) + "automated submission and pre-grading of homework" I homeschooled for a lot of years and my first blog was a homeschooling blog. I was involved with The TAG Project as part of supporting my homeschooling and I really hate this idea. "Teaching to the test" is terrible teaching. Multi-guess answers aren't good ways to test student knowledge. They are just convenient for a teacher who needs to grade 25+ students. Please don't run around thinking of yet more ways to making it nominally more convenient and easy for individual teachers to process tests or homework or whatever from large numbers of students. This is not a way to enhance the transfer of knowledge to the future generation and is also not a good credentialing method. There are cool things being done with tech in the education space. There are people going online and learning things they want to learn because there are rich materials available for free. But they don't look anything like "pre-grading of homework" for teachers.

lgregg 6y ago
Yeah, but you're making an assumption, who said it had to be multi-choice answers? Why can't someone build a NLP model to test essays of middle schoolers or high schoolers graders on American history? You can look for learning objectives in the essay and grade upon that. Why not build a tool with D3 or another modeling tool to teach chemistry? You could have them build compounds and then if they make a mistake, insert via a tool tip and tell them why. As they combine things, you read out what it is or test them based on the string they put in for the compound. You have to build it and name it correctly.
- DoreenMichele 6y ago
  Why not build a tool with D3 or another modeling tool to teach chemistry? You could have them build compounds and then if they make a mistake, insert via a tool tip and tell them why. As they combine things, you read out what it is or test them based on the string they put in for the compound. You have to build it and name it correctly. This sounds really cool and is completely unrelated to what I am saying "Please don't do."
arvidkahl 6y ago

Diesel555 6y ago

I guess that's what I did. Niche is fighter pilot, product is generating target imagery for training. dieselplanning.com.