LawSync.io

LawSync.io

Inspiration

This project began with a conversation with a legal team working with property owners.

They explained that keeping up with local legislative activity in California had become too manual to manage. Their work required awareness of:

Upcoming city council meetings
Proposed or draft legislation related to rent control and fair housing
Rules coming from cities, counties, and unincorporated communities

To stay informed, they had already taken the obvious step — subscribing to many official city and county email newsletters and checking council websites. But the volume quickly became unmanageable.

Every day brought:

Long emails with mixed relevance
Meeting agendas covering many unrelated topics
Repeated updates across jurisdictions
There are few platform that does scouting for the Council, but have limited access to only 2 council and are expensive as they charge $20/month.

Reviewing all of it consistently was taking significant time, and important information was buried among routine announcements.

They asked a simple question:

“Can this be monitored and summarized automatically?”

That question became the starting point for Law Sync Io.

What it does

Law Sync Io automates the work of tracking local legislative activity.

It continuously monitors:

City and county council websites
Upcoming meeting agendas and related documents
Official jurisdiction email notifications

When it detects new or updated items related to housing and landlord regulations, it:

Extracts the relevant information
Summarizes it clearly
Explains the potential impact
Sends an alert so users don’t have to read everything themselves

Instead of reviewing dozens of emails and websites, users receive concise summaries focused only on what matters to them.

For the hackathon, we demonstrated this workflow across five California counties, with the architecture designed to scale statewide.

How we built it

We built Law Sync Io using a coordinated set of AI agents.

At the core is Yoturi.com, which provides browsing, scouting, and deep-research agents capable of navigating complex government websites and detecting meaningful updates.

Our tech stack includes:

Yoturi.com for scouting, browsing, and deep research
AWS Strands for agent creation and orchestration
OpenAI API for reasoning and summarization
React for the frontend
SQL for structured data storage and analytics (just a mockup in sqlite)

Each agent focuses on a specific task — monitoring, filtering, analyzing, or assisting — so the system works continuously without manual effort.

Challenges we identified

After meeting, we create a prototype for scouting, relied on large-scale crawling and retrieval-based methods. While it collected data successfully, it produced too much raw information and struggled with accuracy and relevance.

We learned that firecrawler are only good for static s:

Collecting data is easy; identifying meaningful updates is not
Government websites vary widely in structure
Change detection and filtering are more important than volume

Web-crawler/scraper architecture relies on scheduled scraping of predefined “watch pages,” hashing and diffing content to detect changes, extracting text from pages and attachments, and generating summaries and alerts, but it suffers from lower accuracy due to limited contextual understanding, requires significant ongoing human developer effort to configure and maintain scrapers for changing government sites, and incurs non-trivial infrastructure costs for scheduling, storage, retries, PDF processing, and monitoring to keep the system reliable.

Agent-based browsing improves accuracy because agents:

navigate pages like humans
read context around items
open attachments selectively
reason about “what matters”

Accomplishments that we're proud of

Built an automated system navigate pages like humans, extracts only relevant infromation
Successfully summarized real legislative updates from live sources with citing
Demonstrated a scalable approach to statewide monitoring
Turned an operational burden into a streamlined workflow
Notification for only relevant information on email

This was only possible using the Yutori.com agents as it can work like a hundred of computer browser running parellel to scout the website for all the councils.

All of those can be done just by using the API by Yutori.