On a Saturday in late February, Travis County Fire Chief Ken Bailey sat in front of a computer and ran a simulation. High fire danger day. He plugged in the predicted weather and watched the software model where fire would travel if it started.

"We wanted to run a simulation to see where we were at most risk to have a larger fire," Bailey told KVUE. The system is called Mitigate, developed by fire science company FiSci. It combines vegetation data, weather inputs, and topography to simulate wildfire spread. Travis County is the first county in the nation to adopt it. Four departments now use the platform: Travis County Fire Rescue, Lake Travis Fire Rescue, Pflugerville Fire Department, and Westlake Fire Department.

Bailey identified hot spots. Created a game plan. Sent crews to assess what they'd face if something ignited. A fire chief reading a predictive map on his weekend, planning for something that hasn't happened yet. Preparation in Central Texas now.

The timing of adoption tells its own story. Travis County's burn ban was reinstated February 10 after just two weeks without one. Extended February 24 through March 24. The Southwestern Travis County Groundwater Conservation District declared extreme drought on February 19. "Vegetation across Travis County remains critically dry, and even a small spark can quickly turn into a dangerous wildfire," Chief Fire Marshal Gary Howell said. For context: the 2011 Labor Day fires burned 7,000 acres in Travis County and destroyed 57 homes. In neighboring Bastrop County, 34,000 acres and 1,700 homes. That was fifteen years ago, before the current building boom, before this drought cycle.

Bailey's description of what his department protects is precise: "Central Texas is a patchwork of grasslands, wooded corridors, and dense neighborhoods. Each of those environments reacts differently to fire." That patchwork makes the modeling both useful and insufficient. A simulation models vegetation and wind. The family that tosses a cigarette out a car window on Ranch Road 620 remains outside its frame.

Bobby Abbott, Lake Travis Fire Rescue's chief, framed the tool operationally: "We can estimate how quickly a fire might compromise evacuation routes, how wide the fire front could become, and when sheltering in place or moving residents to a refuge area might actually be safer than a full evacuation." His department has been at this longer than most, having mitigated over 180 acres of vegetation and protected over $400 million in assessed property value. For LTFR, the AI layer sits on top of years of physical brush clearing.

Then there's Pflugerville. Northeast of Austin. Suburban. A place that has never thought of itself as wildfire country. Fire Chief Nick Perkins framed his department's interest differently: "By identifying areas more likely to experience large-scale fire spread, we can focus mitigation efforts, community outreach, and public education where they're most likely to make a difference."

Community outreach. Public education. Perkins is describing the work of convincing people they live somewhere fire can reach. A department absorbing a reality its community hasn't yet accepted.

Austin is expected to double its population in the next 30 years. Much of that growth will push into what fire professionals call the wildland-urban interface, where development shares a back fence with land that was recently pasture or scrub. The households moving into that interface won't all have the same capacity to clear brush, retrofit a roof, or absorb the cost of mitigation the fire department is now asking for.

Bailey also noted something more mundane than fire prediction: "We all have fixed budgets, generally speaking, and so should we staff up today, does that match, because that costs money." A fire chief using AI to justify overtime requests. The daily institutional reality underneath the technology.

No public accounting yet exists of scenarios the model has underestimated, data inputs it lacks for Central Texas conditions, or how it handles human ignition patterns. The departments are comparing projections with past fire behavior and finding close matches, according to a joint release. Early validation looks promising, though a long way from proven.

Bailey can see the risk on his Saturday simulation. He can send crews to assess the hot spots. He can build a game plan. Whether the families filling in the subdivisions behind the tree line will ever look at what he's looking at remains the open question. The screen shows the fire. The audience for it, so far, is the fire station.