Inside the Law School's AI Lab: LeaseChat, Access to Justice, and Practical AI for Lawyers
UChicago Law ran an AI Lab with a straightforward brief: build and ship a legal tool. The result was LeaseChat, a free chatbot that helps renters read leases and spot their rights under local law. With tens of millions of rented homes in the U.S., even modest accuracy and reach can matter at scale. As the Lab's lead, SixFifty CEO Kimball Dean Parker, '13, put it, the potential footprint is huge.
For legal professionals, this wasn't theory. It was product, distribution, and legal risk-under time pressure-run like a start-up.
Student-led, skill-first
Marley McAliley, '27, came from Google PR and owned media outreach. She built a press list, wrote pitches, and secured coverage from NBC Chicago and Telemundo Chicago-turning interest into users.
Alfredo Taboada, LLM '26, joined with minimal AI experience and expected a coding boot camp. Instead, he got a team environment where roles matched strengths: outreach, research, testing, product.
When Parker noticed that Adan Ordonez, '27, could code, he let him run with it. Ordonez built the prototype in days, drawing on self-taught skills and prior tools he'd shipped (like LawBandit, a study app with hundreds of users). His takeaway: if you know the law and you know AI tools, you can work faster and more efficiently.
The technical pivot that mattered
The team debated a classic build decision: plug a curated database into an LLM-or rely on general AI reasoning and invest in stronger prompts. They expected the database route to be more accurate. It wasn't.
As newer models rolled out, hallucination rates dropped, and prompt quality moved the needle more than a homegrown database. The class pivoted to prompt engineering and tight instructions instead of building a massive knowledge base. It saved time and kept output consistent across jurisdictions.
Distribution: partnerships vs. press
The first plan was partnerships-organizations that could push LeaseChat to renters. Cold outreach didn't convert. So the team shifted to a simple marketing plan and earned media.
PR worked. McAliley coordinated on-campus filming and got LeaseChat on air. That created an acquisition flywheel: awareness, traffic, user feedback, improvements.
Risk management: legal info, not legal advice
From day one, the team drew a firm line. LeaseChat offers legal information and points to issues a renter should consider. It doesn't replace counsel, and it says so clearly.
That clarity serves users and reduces risk. It also aligns with tech competence expectations in our profession. The spirit of ABA Model Rule 1.1, Comment 8, applies here-lawyers should keep current with relevant technology.
Why this matters for legal teams
LeaseChat shows what a small, focused team can do with modern LLMs and a tight scope. It also highlights a gap many firms and legal departments have: product thinking.
- Start small: pick one high-volume client issue (leases, NDAs, late fees, habitability) and build a "first pass" assistant that flags risk and explains options.
- Standardize prompts: maintain a shared prompt library, test across cities/states, and log failure cases to improve instructions-not just the model.
- Be explicit on disclaimers: repeat the "information, not advice" message and provide clear escalation paths to a lawyer.
- Decide on distribution: partners are slow; press, content, and internal champions move faster.
- Measure: track accuracy, user satisfaction, and referral-to-counsel rates. Ship updates weekly, not quarterly.
- Upskill the team: short workshops on prompting can raise output quality across the board.
If you want a quick way to level up prompting skills, see this prompt-engineering collection: Prompt Engineering resources.
Student perspectives worth noting
McAliley: AI can widen access, but it won't replace counsel-especially when stakes are high. Her view: meet the tech head on, use it well, and use it responsibly.
Ordonez: confidence grows by building. His class project connected his coding to practical legal use, and he's now pursuing a JD/MBA and writing on AI and attorney-client privilege.
Taboada: the entry barrier is lower than expected. He now sees AI as a way for smaller firms to compete, not a threat reserved for big players.
What's next
Since launch, LeaseChat has analyzed leases daily and helped hundreds of renters. Parker is working with cities and counties to refer residents and exploring a long-term host to keep improving the tool.
The AI Lab returns this fall with a new access-to-justice challenge. Parker noted that students approach AI with a wider lens-they grew up with it-and that perspective pushed the work forward.
Context that frames the opportunity
The U.S. has a massive rental market, which makes lease literacy a public-interest issue with real volume.
Bottom line for legal professionals
Build small tools that reduce friction for clients. Use current LLMs with strong prompts. Be explicit about scope and escalation to counsel.
Do that consistently, and you'll improve service, cut cycle time, and create more room for the judgment calls that matter.
Your membership also unlocks:
