A bunch of startup ideas, go build them or something. May actually be bad ideas, these are all just quick thoughts.
Automated notetaking: Take speech-to-text and recent advances in automated summarization, perhaps peppered with some hand-coded rules around entity recognition and you have an app that automatically takes notes for you.
- Good: People seem to like taking notes
- Bad: Potentially anyone could replicate the core tech? Also maybe it's not enough for this to be 90% good but 99% for adoption. However, even if imperfect maybe it's good anyway if you wouldn't have otherwise taken any notes
SAP, but good. Unbeknown to startupland, the world runs on SAP. SAP needs a replacement, it feels clunky to use, API access not trivial, expensive (From my experience doing consulting, maybe things are different elsewhere idk)
- Good: Match features, but cheaper and nice UI -> win. B2B -> charge more
- Bad: Taking SAP out may be like doing brain or heart surgery to a company, potentially very costly and migration could fail and destroy trust in the product
ML for faster engineering/scientific computation. Use ML to compute approximate solutions in mechanical or fluid simulations, then use regular solver to get solution. What took a day maybe could take minutes?
- Good: Maybe be doable as a plugin to regular industry standard (e.g. ANSYS') software to ease adoption. Very niche sector, not much competition. Maybe there is room for faster software build on top of Rust+GPUs
- Bad: Cost of training the model can be high?
Branded houses. We use brands to distinguish quality, yet landlording is still a fragmented profession. Agencies do not fully solve the problem and are still local. Imagine Hilton houses so when you go to a new city to live in there instead of using the local zillow/rightmove/idealista you just look at the nice brands and any house will do.
- Good: Possible trend towards more mobility, less attachment to long-term ownership of houses, this plays into that? People will pay for certainty rather than dealing with a new individual landlord or agency
- Bad: capital costs to buy the houses. Though it can be started small then scale
Tools for measuring developer productivity. Software engineering doesn't seem to be very empirical, let's change that. Should a team do more or less testing? Switch to a different language? Use this or that way of structuring code? What doesn't get measured doesn't get improved and what is hard to measure doesn't get measured. So imagine a time tracker+text input well integrated with IDEs that allows devs to say why things took that long, issues they found, etc. This way it can be noticed, in the long run, what is detracting time, what sort of issues occur all the time; this helps prioritise. Think of it as Rescuetime+JIRA+Some NLP
- Good: Low friction, can save lots of engineering-hours/$
- Bad: Some people don't like to be measured in such way which may difficult adoption.
A global trip optimizer that is -nice-. Google flights/Skyscanner work great for flights, booking.com/etc work great for hotel. Very hard to find a website that has a great UI to search specific flight options+hotels
- Good: Proven model, maybe can be just sustained with ads or % of fee payed a la Trainline
- Bad: Maybe this already exists?
Public transit cards, done well. Collecting transit cards from cities all over the world is part of the rootless cosmopolitan's life. But it doesn't have to be like that: In Japan a single card allows the use of virtually all transit. An oyster card should work in Edinburgh, or in New York (Or they can just be made contactless). Create a system that's easy to install (A contactless reader and associated hardware) that interfaces with legacy systems and uses contactless cards and/or others, sell the service to transit operators, give them nice dashboards
- Good: For the users perspective, no more paper tickets or weird quirky things that depend on each city
- Bad: Working with lots of agencies, idk if they would have an incentive to buy this
VR office. Being able to take pdfs out of your screen and put them around in a room, pinch to zoom on them, spatially rearrange content etc. Effectively more screen space+immersion.
- Good: Wow factor, useful
- Bad: Needs people to buy an Oculus or related. Does this have demand? Dizziness? Not enough resolution to read things in VR?
A prediction market for statups. Let people bet on who will fail, who will succeed and when. Generates information, lets VC scout good pickers.
- Good: Sounds easy to do, just do a nice UI, sit in the middle and collect fees.
- Bad: May have legal issues depending on jurisdiction