The Filing Cabinet Problem
For forty years, agencies built repositories of institutional knowledge that were human-readable but machine-invisible. LLMs changed that. Here's what it means for agencies that act first.
ReadBeacon applies AI to the millions of documents transportation agencies already own - turning buried institutional knowledge into source-linked answers across design, construction, and operations.
Transportation agencies run on documents. Beacon makes them answer questions — across every phase of the project lifecycle.
Project managers spend weeks emailing eight offices for level-of-effort hours — starting from scratch every time, even when the agency has done the same work a hundred times before.
Active construction programs generate thousands of inspection reports, submittals, and change orders. Finding what's relevant — across systems, across years — shouldn't take a week.
TMC operators manage incidents under pressure with 600-page SOPs and contact lists spread across spreadsheets. Every second of confusion is seconds of lane closure.
Large DOT fleets run at 70% camera uptime. The other 30% is invisible breakage accumulating across processes no one has mapped. IP mismatches, iNet sync failures, stream manager demand spikes — classified and ticketed.
Every DOT is sitting on millions of documents it can file but can't read. Beacon changes what those documents can do.
Beacon sits on top of the files you already have — and turns them into something your people can actually use.
No migration. No rip-and-replace. Beacon connects to what's already there — and makes it answerable.
Plugs into ProjectWise, SharePoint, shared drives, and meeting archives. Vendor-agnostic. No data migration required.
AI extracts structured attributes from unstructured files — dates, locations, asset IDs, compliance status, project numbers.
Cross-references across millions of documents, connecting related records even when they were filed in different systems a decade apart.
Delivers source-linked answers through purpose-built interfaces. Every response cites its document. Every query leaves an audit trail.
It's easy to stand something up. It's hard to make it trusted.
Consultants and in-house teams can build apps. We build decision systems you can defend.
"The cost of software is going to zero.
The cost of trust in what has been built is not."
I can show my work. Every answer traces to its source — document, section, paragraph. Human judgment stays in the loop.
I know when to trust it. Every answer carries a confidence score. Test sets run daily. Error rates tracked and published.
I can defend it. Every query is logged and reviewable. FOIA-ready. PE-reviewable. No black box.
TMC-React is live with operators on overnight shifts — not just in training. Estimator is rolling out across the agency's concept-development program, drawing on a 10.5-million-document ProjectWise environment.
Projects in GA, NC, and SD — scoping engagements in six additional states.
On top of the core ATMS deployment, Beacon built an AI Runbook from over 300 recorded meetings — every incident, diagnosis, and resolution from the most operationally complex system the agency runs. TMC-React went live with operators across the state. Estimator is now rolling out across the agency's concept-development program, drawing on a 10.5-million-document, 21.5-terabyte ProjectWise environment.
The ATMS AI Runbook and TMC-React together mean that when something breaks at 2am, the answer isn't in a binder anymore.
The agency's Deputy Chief Engineer cited projects taking seven years from concept to construction for a single roundabout. The estimation bottleneck isn't expensive in direct dollars — it holds up everything downstream. Estimator is the first AI platform deployed inside a state DOT for concept-development estimation.
Most questions about Beacon start with what it isn't. We'll save you the guessing.
Beacon is vendor-agnostic and overlays existing systems via API. It doesn't touch your field devices or replace your traffic management infrastructure. It makes what you already have smarter.
A search engine finds files. Beacon reads them. It extracts structured intelligence — dates, compliance status, project relationships — and delivers answers with source citations, not a list of documents to click through.
Pilot sites are live in 4–6 weeks. Beacon connects to your existing repositories on day one. No integration sprint. No steering committee. Operators use the product on day two.
Beacon was founded by a civil engineer who spent a decade delivering major infrastructure programs at CH2M Hill before moving into executive roles at Oracle and EarthLink. The company is built on the premise that the gap isn't technology — it's translation between how government agencies work and how modern AI is deployed.
Our advisors include former state DOT directors, senior engineers who have run billion-dollar programs, and digital leaders from the world's largest infrastructure firms. We get invited into rooms that most startups never reach because we speak the language of the people in them.
B.S. and M.S. in Civil Engineering, Georgia Tech. Decade at CH2M Hill. Executive roles at Oracle and EarthLink. Bloomberg and CNBC commentator. Keynote speaker at AASHTO CTE.