Requests For Startups
What to build to make the future better
The ability to split-test different AI models
Many AI models exist: ChatGPT, Gemini, Claude, and hundreds more.
There should be a way with one line of code to split test model performance based on speed, accuracy, and user feedback… So companies can ensure they are using the best model for their use case.
Modernizing police databases and database enrichment
Imagine if every time a police officer pulled someone over they could see basic public information such as where they work, their family details, and more.
This data could be crucial to preventing cases of human trafficking.
For example, if someone is hundreds of miles away from their place of work on a Monday and claims the child in their back seat is theirs, but their “family Christmas photo” does not show that child, this data could be crucial to prevent human trafficking.
Police Software in General
Everyone has shied away from building software for law enforcement
Not because it is hard— but because it has been socially unacceptable.
We see this with the political movements “Defund the Police” and “Abolish the Police”.
These people are wrong.
The way to fix our broken law enforcement is to build more technology.
This will allow current enforcement agencies to be more efficient and transparent.
There are hundreds of billion-dollar startups to be built here.
Missiles
Whenever a decades-old military system becomes synonymous with a function, such as “Stinger” missiles with surface-to-air anti-aircraft capabilities, there is likely an opportunity to make it cheaper and better.
Stinger missiles have been a crucial part of the US arsenal since the 1970s and have proven a critical defense weapon for militaries and peacekeeping organizations.
But there’s one minor problem… these missiles cost $120,000 to $150,000 per missile, with costs expected to rise 50% due to increased demand in the next few years.
Similar to Stinger missiles, “Javelin” anti-tank missiles first invented in the 1980s cost around $200,000 to $240,000 per missile— with increasing demand as well.
There’s some opportunity here for someone who does not shy away from building military technology to build missiles with better capabilities than our current Javelin and Stinger arsenal.
Government Spending Software
Local NGOs and organizations spend hundreds of billions of dollars every year.
But most of this money disappears into the pockets of NGOs run by politicians’ friends.
The “impact” of most NGOs is just renting a few vans, handing out the occasional sandwich, and collecting multi-million dollar contracts.
No one truly knows if this “investment” is really effective in combating drugs, violence, and homelessness in our inner cities.
We need a centralized system for social organizations to log their activity into…
With logged salaries, services rendered, and the specific people helped.
Think of a “CRM” but those in most need— that can be used to track government and NGO organizations and their true effectiveness.
Data products are powerful at scale with large datasets.
Interestingly, the organization with the largest datasets to ever exist, the US government, doesn’t seem to be using their data at all.
The product is buildable.
But sales is the bottleneck— you’d have to start mobilizing the voting population of small local governments and expand from there.
Bidding system for inventory
For highly in-demand products, like Nvidia GPUs, there should be a marketplace with buy-and-sell orders instead of fixed prices.
Virtual Bio Labs
Biotech innovation is held back by iteration speed.
For software, you can deploy code and get feedback immediately. But running a single experiment for biotech can take weeks, months, or years. We need the ability to run biotech experiments “in the cloud”.
Software to simulate the entire human body would be great!
But simulating every cell interaction is beyond current computing power.
It would be great to have labs run completely by robots, where you can control the robots in the lab to run experiments with code.
The Ethereum network built Solidity for crypto-specific use cases.
We need a similar programming language or unified API for Biotech.
Need to run an experiment?
Enter a few lines of code, hit submit, and the instructions for your experiment will be sent to robotic labs.
A robot will follow your instructions, conduct your experiment, and send you the results.
This would create a world where anyone can run biotech experiments, speed up iteration speed, and lead to radically more biotech innovation.
Challenges? This isn’t strictly a software problem of creating a new programming language, you need the software to control hardware.
You would have to significantly augment current machines or even completely build new biotech machines from scratch.
Initially, you could work with people with co-working lab memberships to run the experiments manually.
It's a hard problem with massive payoff potential for Humanity.
Nutrigenomics
There are ethnic and racial differences between humans.
Tibetans have evolved for high-elevation living, East Asians have evolved to be lactose intolerant, and Dutch people have evolved to be tall.
These differences are undeniable, but since the Nazi’s defeat in 1945 no one has wanted to touch genetics— it has been too controversial.
But to deny these differences between humans is immensely harmful.
Is it crazy to say different people who evolved on different continents with different local food sources have evolved to process foods differently?
Probably not.
If your ancestors lived in Portugal or the Chilean mountains for tens of thousands of years wouldn’t you have evolved to better process local foods from these regions?
In short— there is a huge opportunity to personalize nutrition according to genetic heritage.
But the challenge in Nutrigenomics isn’t standing up to the mob and saying the controversial— it is also a feat of science to do the technical genetics research.
All-in-one AI Sales Software
Everyone talks about “compound startups” with centralized data such as Rippling in HR Tech…
But no one has built a compound startup in the sales space.
CRM, AI Agents, Dialer, and Sequencing all in one place.
Bus tracking software for school districts
Ability to track pickups, dropoffs, and live mapping for parents.
Video/Audio Translation
Only 18% of the world speaks English.
Large YouTube channels like Mr Beast and major movie producers spend hundreds of thousands of dollars per year translating their media.
There is an opportunity to auto-translate and dub videos, movies, and more with software.
It could also grow organically allowing everyday people to translate a YouTube video for free.
AI Created Docs & User Onboarding
Most API products and software companies have documentation for their products that need to be updated every time they update their product.
Instead, there should be an integration into their code base that then updates their documentation and onboarding experiences automatically for every product update.
This would remove tons of tedious work.
AI Data Extraction
Today, all large data startups are created through complex HTML regex and JSON intercepts.
Engineers must constantly monitor web structure changes and spend hours searching through JSON intercept structures for each site when building datasets.
The ability to plug any site or dataset origin and automatically extract data would save thousands of engineering hours.
Genetically Altered Chickens
1/2 of all chicks are born male, as a result— 260,000,000 chicks are euthanized per year in the USA alone.
In Europe they are scanning eggs before hatching to sort out male chickens… but a better solution is to make male chickens only produce X chromosomes, or even at a higher rate.
This would significantly increase the ROI for farmers and prevent billions of chicken lives from slaughter globally
Book Writing AI Platform For Authors
Today, all authors use Google Docs or Microsoft Word to write their books.
There must be a better way to keep track of plots, subplots, and the progressions of stories.
In the same way that programmers have IDEs, authors need BREs— book writing environments.
AI Account Manager & Customer Success Managers
With the rise of AI, hundreds of AI sales software and “AI SDR” startups have popped up.
These startups are all focused on generating revenue from new customers.
In addition to AI sales startups…
There are late stage startups like Zendesk and killer upstarts like Pylon, focused on customers service and solving customers issues.
Uniquely, I believe there is a position for AI account manager and customer success startups— that track how customers interact with your product and cater to them.
Increasing customer satisfaction, reducing churn, and increasing upsells/NRR.
This could be done through a ton of ways such as tracking clicks and product usage of customers— then scheduling calls with customers or sending them a product how-to guide whenever they encounter product friction…
Or even better, when customer product usage indicates interest in premium features $$$
A platform that lets anyone create and manage custom relationship contracts that replace traditional marriage licenses and relationship NDAs.
The Government has no place in defining marriage as a legal status.
Think "LegalZoom for relationships" - handling everything from disclosures and property rights to healthcare decisions through smart legal agreements that work with existing systems.
With 2M+ marriages annually in the US and growing demand for prenups, there’s a $100B+ market of people who want to define relationships on their own terms.
Cursor for sales teams.
Control your sales CRM and outreach with natural language.
Warm CRM Linking.
Many founders and sales teams have a multitude of potential leads in their Instagram, LinkedIn, and X followings… but for some reason this data is not added to CRMs automatically.
If a potential customer likes your tweets or follows you, don’t you want this data logged automatically?