Suggested by 
Eric Niehaus
almost 2 years ago
Problem
99% of encounters regarding the law — with law enforcement and other citizens — happen before anything goes to trial or lawyers are involved. But no citizens knows all the laws or all their rights, especially small, local laws and ordinances. This can lead to rights being infringed and freedoms being limited, especially when confronting law enforcement.
We have many rights. But those can't manifest unless they're known. Waiting to go to trial is silly, especially since it's often expensive, risky, and time-consuming. 99% never go to court. There is a very large pre-trial market for legal resolution.
The big problem is the little guy not having access to the knowledge of his rights in-the-moment, where most "legal issues" are settled, albeit informally.
Examples
The value here is that an AI can scour and synthesize the heap of laws in ways we can't, so it has as many applications as there are laws
Solution
AI Attorney — a lawyer in your pocket for the 99% of the pre-trial/lawyer legal encounters. Tell it where you are and what's happening and it'll be able to provide every party's rights and the correct and legal resolution to the situation, accounting for local, state, and federal laws, providing specific citations. It could also provide a "Next Steps" menu, e.g. if the cop says the law isn't real, it'll advise the citizen what to do next.
Business Model
Not sure exactly. But it should be free. The conversation among friends would be something like "You should really have AttorneyAI downloaded on your phone, just in case. It saved me when I was getting a ticket the other day."
Competition
Most apps of this kind merely record the encounter but do nothing to equip citizens with in-the-moment knowledge of their rights, e.g. the LegalEqualizer App (no longer available)
Are you interested in addressing this Unmet Need?
Founder @ Auril AI/ Earshot
Christina is right - LLMs left to their own devices are likely to hallucinate and make up their own case law.
To put more color to what Christina said about the "Ask Jeeves" approach, the modern LLM version of that implementation would likely be a RAG-driven architecture based on relevant case law. Such an architecture could serve up targeted legal advice by finding and summarizing case law based on contextual information it receives from an app.
It's worth noting though, performant & stable production RAG is one of the harder challenges in Generative AI right now, and many firms just bail on RAG entirely and try to pass tons and tons of information into model prompts (an expensive and limited workaround). You'd want to have a savvy technical expert working on it in order to get something performant enough to serve up legal advice, even with a disclaimer that advice shouldn't replace reaching out to a lawyer.
That model could also then pilot the user through a collaborative search to match them with attorneys, per Jeremy's suggestion.
An interesting idea!
If you need any help sketching out what this AI Architecture would need to look like, I'd be more than happy to pitch in a little time to help you get started. My team and I are building apps like this all the time.
CEO | Founder | Managing Partner @ Platform Venture Studio
The business model here could be as a lead-gen engine for criminal attorneys. The app could triage the situation and then match to available criminal attorneys. I imagine that most people who end up getting arrested don't have an existing criminal attorney relationship.
AI Strategist @ Geeky Insights
AI has been faking case law. I'm not sure if GenAI is there just yet for legal cases. What about an FAQ app like Ask Jeeves that curates responses based on keywords. A bit safer to implement.