Tyler Griffin's bots follow up on quotes, watch reviews, pull numbers and work inside ServiceTitan. One built boiler quotes after checking supplier pricing. Another found the closest tech for a recall and worked with his office to book it.
AI for quotes and estimates
32 real stories from businesses using AI for this job. Newest first.
Extract scope requirements and open questions from a lighting specification before putting a price on the job.
Separate price-book jobs from custom work before drafting a garage-door quote.
Keep existing snow customers and new quote requests in separate storm-day queues.
Check the age and completeness of site measurements before using them in a landscape or snow bid.
A new caller said ChatGPT sent him. Then he brought up the percentage the owner plans to pay cleaners. The owner admits it may have been a lucky guess, but it's a good reminder about what you type into a chatbot.
Cross-check a verified roof report against site notes before drafting the material list.
Pay-by-link invoices, estimates customers approve online, and one dispatcher screen made the difference. Customers still preferred to call or text.
"Someone asking for help shouldn't get homework back." No more asking for sizes, joists or photos up front.
It triages email, drafts quotes and contracts, and builds the crew schedule. His rule: "it prepares, I approve."
Document the rules in a pricing spreadsheet before turning it into a simple quote-calculator demo.
After testing on more than 100 takeoffs, he says it gets to 90-95% right. Others say it needs clean PDFs, setup time, and a human who still knows the job.
He enters his time and expenses; ChatGPT splits the bill into 10 cost buckets. He says it saves him half a day to a full day a week.
One contractor did her aunt's job at cost and ChatGPT still called it overpriced. A Phoenix GC says spelling out the specs is what holds up.
She fed it SOPs, example estimates and hundreds of old emails. Before that, fixing its work took longer than doing it by hand.
Messy site notes go in, a clean quote in their format comes out, and invoices finally go out on time.
Quotes that took him an hour per 10 now come out with datasheets in about 15 minutes.
AI is weakest reading dense drawings as images. Pull out the notes and schedules as text first.
It got there only because he gave it every variable. Others report about 70% accuracy on AI takeoffs.
A skeptic's view: entering everything into the AI is the same work as writing the quote.
One uses an estimating tool with built-in AI takeoff that saves him hours, but warns never to replace estimator instincts.
Showing the customer their own yard finished beat 'we'll think about it.'
Answering calls, quote follow-ups, rebooking past customers and review requests, all automated.
An evening of building, and now the part that took him hours runs in 30 seconds.
"9/10 times the slop comes up with something wrong." The tasks that did work: counting simple items, summarizing narratives, comparing drawing changes.
He held his price, took $70 off, and says online "fair price" estimates were far too low even before AI.
He hated the awkward "did you decide?" call, so he had ChatGPT write a short follow-up text for every quote.
A ChatGPT-written description on every line, e-signature, and automatic nudges at 3, 7 and 10 days.
A structured "experienced electrical estimator" prompt turns drawings into a draft takeoff.
Organize verified home and equipment details into a pre-assessment packet for a qualified electrician, not an AI approval.
The client gets a detailed recap email 5 minutes after the visit, and the proposal comes from the same transcript.
The company's Copilot agent sorted items into in scope, maybe in scope and unclear. A "neat party trick," he says, useful for a final check.
Want this running in your business without the DIY? Rise Systems in Oconomowoc sets these up.
Free 15-minute AI check-up