Process thousands of utility permit emails with AI agents
Freedom Forever
What they did with AI
Freedom Forever, which handles about 5,000 residential solar permits a month, built AI agents to read and act on the flood of utility and municipal permit emails (inspection notices, payment requirements, correction requests, approvals). The agent processed about 19,000 permit emails in its first month and fully handled about 71% without a human, at $0.20-$0.35 per run. A backfill turned up emails that had sat unprocessed for an average of three weeks, which got stalled projects moving again. A second agent validates uploaded utility bills, including photos of paper bills.
Results above are as reported by the vendor or company in the source and haven't been independently audited by this site. Listing a tool is not an endorsement, and nobody paid for this entry.
Related use cases
Build solar proposals in under 10 minutes and cut change orders
Mpower Solar
New York City installer Mpower Solar uses AI-assisted design to build proposals for tricky canopy and tilt-mounted rooftops and adjust them live in meetings. Aurora reports proposal creation went from about 30 minutes to under 10, the change-order rate dropped from 50% to under 10%, and site surveys went from 2 hours to 30-60 minutes. The case study also says consistent design documentation led to fewer questions on submissions to state agencies and utilities.
Run 27 back-office AI agents with strict approval guardrails
Lake Stevens Pete (owner of an HVAC company and other businesses)
An owner of several small businesses, including an HVAC company, counted 27 active AI agents and 36 routines (26 on a schedule) in use across his companies. They pull statements, file invoices, update sales and cost sheets, flag overdue maintenance and track renewal dates. His rules: only three emails ever send on their own, everything else is a draft he approves, and nothing posts publicly or pays a bill without his yes. He keeps an inventory of all agents, which caught a failed reminder and two routines doing the same job.
Compare submittals against specs with an AI agent
Haskell
On its own Jacksonville HQ renovation, design-builder Haskell posted QR codes around the jobsite so workers and trade partners could ask an AI agent about specs, submittals and drawings and get answers with links to the source documents. Assistant PMs also use a templated agent that compares a trade partner's submittal to the spec and lays out variances in a side-by-side table. Haskell estimates it saves about 10 minutes per field question and more than an hour per submittal review.