How AI powers DisasterChief

When a tenant submits a request — by text or via the web form — DisasterChief passes the full description to AI for analysis. It classifies the issue into one of seven trade categories (plumbing, electrical, HVAC, appliances, structural, pest control, or general) and assigns an urgency score from 1 to 5.

High-urgency results (4–5) trigger an immediate owner or manager alert before technician outreach even begins. The category determines which technicians are contacted and in what order.

Every categorization is logged with a confidence score. If you ever correct the AI's assessment in your dashboard, that correction is saved as a training example — the system gets more accurate for your specific request types over time.

If AI can't categorize a request with enough confidence — usually because the description is too vague — it generates a targeted follow-up question instead of falling back to "general." The question is specific to the description, not a generic prompt.

Tenant-submitted requests: The follow-up is sent to the tenant via SMS. Technician outreach is paused until they reply. Their answer is combined with the original description, and the full pipeline re-runs with the enriched input.

Owner-submitted requests (from the dashboard): The follow-up question appears inline in the request form — no SMS involved. The owner's answer is folded back in before routing begins.

This dramatically reduces the number of requests that land in the "general" bucket, which means better technician matching and fewer misdirected jobs.

Every outreach message sent to a technician is drafted by AI — not assembled from a generic template. The message includes the actual issue description, property address, unit number, and the name of the owner or management company the request is on behalf of.

The result reads like a real, professional message from you — because it is, just written faster than you could type it.

For service providers (companies with dispatchers), the message is adapted to address a dispatcher rather than an individual technician, with appropriate language for routing work internally.

When a technician submits a quote, DisasterChief compares it against historical quotes for the same trade category and zip code. If the amount is meaningfully above or below the typical range, AI generates a contextual note that appears alongside the quote in your approval request.

The note is specific — "this quote is on the higher end for HVAC work in this area" — rather than a generic flag. It gives you real context before you approve or reject.

The comparison set grows automatically as more jobs are completed through the platform. Early on it will have limited data; over time it becomes a meaningful reference point for your specific market.

When the system needs to hand a situation back to you — no available technician, AI couldn't categorize the request, all providers declined — the alert includes the specific issue, the unit, a summary of what was tried, and a suggested next action.

You won't receive a message that says "no technician found." You'll receive something like: "No licensed electrician was available for Unit 4B's panel issue at 215 Oak St. You may want to contact [your usual electrician] directly or expand your provider list."

The difference matters at 11pm when a tenant's heat is out.

After a job is marked complete, the tenant receives a follow-up text confirming the work and asking if the issue is resolved. This message is written by AI using the actual issue description — not a generic "your request has been completed" prompt.

A tenant who reported a dripping faucet gets a message about their faucet. A tenant who reported a broken heater gets a message about their heat. The specificity gets better response rates and keeps tenants from feeling like they're interacting with a generic ticketing system.

If the tenant confirms resolution (or doesn't respond within 24 hours), the request closes automatically.