511Now

The problem, the night before

The night before a mountain hike I was bouncing between four tabs: wildfire smoke on one site, the weather forecast on another, highway traffic on a third, and the mountain pass cameras on a fourth. The information all existed, and all of it was official. It just wasn’t in one place.

That’s a findability problem, not a data problem — and it looked like exactly the kind of distance AI-assisted development should be able to collapse. So that night I researched the idea and the available domains, bought 511now.com, and started building with Grok Bot and several other AI tools. By the next day the site was live.

What shipped

One place for the Northwest’s go/no-go questions, across Washington, Oregon and Idaho:

• Wildfire perimeters from NIFC, with acreage and containment, nearest first
• Air quality and smoke — official AQI readings ranked worst-first, with “how far can you see” cross-checked against NWS visibility reports
• Weather and NWS alerts for passes and towns
• Traffic incidents and closures from the state DOTs
• 2,700+ official camera feeds — WSDOT, ODOT, Idaho 511 and the National Park Service — browsable by road, by region, or by “near me”
• A unified map layering fires, traffic, alerts and cameras, with context per camera: how far the nearest wildfire perimeter is, and whether a weather alert covers it

The build itself is deliberately boring: static pages, vanilla JavaScript and Leaflet pulling official public feeds. Nothing about the problem needed a framework, and a site people check on a phone at a trailhead needs to be fast more than it needs to be clever.

Aggregation has rules

An aggregator earns trust by being honest about what it is, so the constraints went in on day one:

• Official sources only, credited. Camera stills are hotlinked from published agency URLs, never re-hosted, and a sources page names every agency and what it owns.
• Pictures, not verdicts. The site never guesses a closure or chain requirement from a photo. Every answer page starts by naming the official DOT page as the final authority — then shows you the road.
• No fake liveness. Stills aren’t labelled LIVE unless the camera has a real video stream, and stale data is treated as worse than no data.
• Not an alert system, and says so. Prominent disclaimers separate it from 911, Wireless Emergency Alerts and county emergency management.

The knowledge layer

The part that carries my fingerprints most is the Answers section: curated pages for the questions people actually ask — “Is Snoqualmie open?”, “Do I need chains?”, “Can you see Mount Rainier today?” — each structured as source-first guidance with the relevant cameras walked in driving order. That’s demand-driven knowledge with named authorities: the same discipline I ran at enterprise scale, applied to a consumer site.

Where it stands

The core product is live and free. I’m refining the mobile experience, stress-testing the data feeds against real conditions, and adjusting to what actual visitors check first.

The larger point is what the build demonstrates. AI didn’t invent the problem or supply the judgment — the sourcing rules, the disclaimers and the answer structure are editorial decisions. What AI did was collapse the distance between identifying a real problem and shipping something useful: from four open tabs to a working product in under a day.