10decodersFrom data silos to instant answers.
This client helps state transportation departments, city planners, and public safety agencies manage roads and plan smarter cities. 10Decoders built the AI layer that lets their users ask questions in plain English and get accurate answers instantly — no technical query language, no waiting on a data team.
Overview
Specialized talent, without the overhead
The client's data covers crash records, traffic volumes, road layouts, and pedestrian and cyclist infrastructure — valuable information that's often too technical for non-technical decision-makers to access directly. Their customers needed fast answers to guide real decisions, but the data was scattered across systems and formats, with context buried inside reports never built to be searched.
Every question from a non-technical user landed on an already-stretched analyst's desk. 10Decoders partnered with them to close that gap — building a faster path to their data without compromising security or accuracy.
Under the Hood
Built on a modern AI stack
Hosted entirely inside the client's own cloud environment — no third-party infrastructure.
Models relationships between records, not just rows in a table.
Retrieves and reasons over unstructured reports and documents.
Every answer respects who's asking, automatically.
One platform, safely separated by department and agency.
Conversations keep their context, question after question.
Good decisions were waiting on someone else's bandwidth.
The Challenge
- Crash, traffic, and infrastructure data scattered across systems and formats
- Context locked inside reports that were never built to be searched
- Every non-technical question landed on an already-stretched analyst's desk
- Decision-makers needed fast answers without a technical query language
- Speed couldn't come at the cost of security or accuracy
Our Approach
- Unified transportation records, safety data, maps, reports, and documents in one place
- Built a graph-based agentic RAG layer that understands unstructured context
- Gave users a plain-English conversational interface — no dashboards, no queries
- Freed analysts from repetitive requests to focus on higher-value work
- Layered in role-based access so sensitive data stays protected automatically
The Approach
The Solution
10Decoders designed and built an AI-powered knowledge layer that lets this platform's users ask questions about transportation and safety data in plain English, and get accurate, instant answers — no technical query language, no waiting on a data team.
Day to day, it feels like asking a knowledgeable colleague. Type a question, get an answer. The assistant unifies transportation records, safety data, maps, reports, and documents in one place, while sensitive information stays automatically protected.
How It Fits Together
Records, Maps & Reports
Crash data, traffic volumes, road layouts, and documents — wherever they already live.
10Decoders Assistant
Graph-based agentic RAG, memory, and RBAC tuned to this platform's data.
Planners, Analysts & Officials
Plain-English answers, delivered instantly to the person who asked.
Why It Works
Built for trust, not just speed
Graph-based agentic RAG
Understands relationships across unstructured reports and documents, not just keyword matches.
Memory across sessions
Conversations carry context, so users don't have to repeat themselves question after question.
Role-based access control
Every answer respects who's asking, so sensitive records stay protected automatically.
Multi-tenant by design
One platform safely serves every department and agency without data bleeding across lines.
One unified knowledge layer
Crash records, maps, reports, and documents live in a single searchable place.
Built for non-technical users
No query language and no dashboard to learn — just a question, typed in plain English.
What We Built
Purpose-built for geospatial intelligence
Conversational Query Assistant
A chat interface where planners and analysts ask questions in plain English and get instant, sourced answers.
Graph-Based Agentic RAG Engine
Reasons across unstructured reports and documents using a graph database, not flat keyword search.
Memory & Multi-Tenancy Layer
Conversations retain context across sessions, with each agency or department kept securely separate.
Role-Based Access Control
Sensitive records stay protected automatically, with no added steps for the person asking.
The Shift
What Changed
Anyone can ask, in their own words
Natural language replaced technical queries.
One front door to everything
Records, maps, reports, and documents — reachable through a single conversation.
Answers in seconds, not days
Manual research now happens in real time.
Security stays intact
Sensitive data is protected automatically, with no extra effort.
Independence over dependency
Business users explore data themselves.
Built to grow
The foundation absorbs new datasets as the platform expands.
Delivery, Cost & IP
Built right, owned outright
Delivery Quality
- Delivered by a dedicated 4-person team: 1 architect, 2 AI engineers, 1 full-stack developer
- Built and deployed directly inside the client's own AWS environment
- Positive client appreciation received during in-person delivery reviews
Cost & ROI
- A lean, dedicated team scoped to the engagement — not a generic bench
- Six-month build, from architecture through production-ready delivery
- Graph-RAG foundation is reusable, lowering the cost of adding new datasets
IP & Security
- Full source code handed over to the client
- Role-based access control protects sensitive records automatically
- Hosted entirely within the client's own cloud environment — no vendor lock-in
The Outcome
The Impact
From question to decision in one sitting
They move straight from question to decision, not one week later.
Time back for real analysis
Less time fielding requests, more time on work that needs their expertise.
Data on their own terms
Stakeholders explore the data directly, without routing through anyone else.
