The situation
Ryan Hanus runs Firsthand Lawns & Landscaping out of Orlando, FL. When we started working together, the business was growing fast — 600+ active clients, field crews running daily assessments, and an inbox that never stopped. The problem wasn't leads or demand. It was the back office.
After every assessment, someone had to turn field notes into a quote. Every new inquiry needed a personal reply. Routes had to be planned. Reports had to be pulled. Upsell opportunities were sitting in the CRM going nowhere because nobody had time to act on them. Ryan was the bottleneck in his own shop.
He was doing 10–14 assessments a day and spending his evenings writing quotes. It was unsustainable at the scale he was heading toward.
"We did 14 assessments yesterday. By the time I sat down with my coffee, 11 draft quotes were already in my Jobber queue. $38,000 of work. Drafted overnight."— Ryan Hanus, Founder · Firsthand Lawns & Landscaping · Orlando, FL
Our approach
The first thing we did was spend time in the actual workflow — not a whiteboard session, but watching how the team operated. What data lived where. How leads came in. How quotes got written. What a foreman actually needed before heading to a job site.
Two things became clear immediately. First, the work wasn't complex — it was repetitive. Every quote followed the same logic. Every lead reply followed the same tone. Every morning route could be optimized with the same inputs. Second, the existing stack (Jobber, CompanyCam) already contained everything needed. There was no data migration problem. There was just no automation layer connecting it all.
Our approach: build a layer of specialized AI agents, each owning one job, all running in the background. Nothing auto-sends to a customer. Nothing changes for the crew. Everything surfaces as a draft — Ryan reviews, edits if needed, sends what's right. The goal was to compress three hours of evening admin into a 20-minute morning review.
The best automation doesn't replace judgment — it eliminates the time spent on work that didn't need judgment in the first place.
What we built
We built 11 purpose-built agents, each running on its own trigger and schedule. Some fire on real-time webhooks. Others run on a daily cadence overnight. All output drafts — never auto-actions.
The integration layer
All 11 agents connect to Firsthand's existing stack — Jobber for CRM and job management, CompanyCam for field photos and assessment notes, and Claude as the AI backbone for drafting. No new apps for the crew. No data migration. The agents read where the data already lives and write outputs back into the same tools Ryan already uses.
| Layer | Technology | Role |
|---|---|---|
| CRM & Job Management | Jobber | Source of truth for clients, jobs, quotes, requests |
| Field Assessment | CompanyCam | Photos and notes that trigger quote drafting |
| AI Drafting | Claude (Anthropic) | Reads context, drafts replies, quotes, and reports in Ryan's voice |
| Event Triggers | Webhooks + Zapier | Real-time firing on job events, inquiry submissions, quote status changes |
| Crew Comms | SMS | Personalized morning briefings per foreman |
Results
The numbers from production tell the story better than we can:
- $38,200 in quotes drafted from a single morning's batch — 14 assessments, 11 draft quotes ready before coffee
- 30 seconds from assessment-complete to draft quote appearing in Jobber
- 600+ active clients now managed by the agent stack with no additional admin headcount
- $27,000 in annual recurring revenue uplift from the upsell engine surfacing maintenance pricing opportunities that were previously missed
- 3 hours of evening admin compressed to a 20-minute morning review
The shift that mattered most wasn't the time saved — it was where Ryan's attention went instead. With the back office running itself, he could focus on sales, hiring, and the strategic decisions that actually move the business. The agents didn't replace his judgment. They eliminated everything that didn't need it.
From client project to product
What started as a custom automation stack for Firsthand Lawns became the foundation for something larger. The agent architecture we built was repeatable across any field service trade running on Jobber, ServiceTitan, or Housecall Pro. Same problems. Same data sources. Same opportunity.
We spun the system into Drafted — a standalone product now onboarding HVAC, plumbing, electrical, roofing, pool service, and pest control operators. Firsthand Lawns remains Customer Zero, still running the full 11-agent stack in production every day.
Firsthand Lawns didn't just get a back office automation — they became the proof that the model works at scale. Every operator we onboard after them is running a system battle-tested at 600+ clients and 14 assessments a day.
What we learned
- Drafts, not actions. The most important design decision was making every agent output a draft, never an auto-send. This removes the fear of automation for operators and keeps humans in the loop where it matters — customer communication and money.
- The stack is the moat. Jobber and CompanyCam already had the data. The value we added was the layer that connected it, read it intelligently, and surfaced outputs in the right place at the right time.
- One job per agent. Eleven specialized agents outperform one general-purpose assistant every time. Each agent is easier to test, easier to improve, and easier for the operator to trust.
- The back office was the bottleneck, not the business. Ryan had demand. He had crews. He had clients. The constraint was admin — and once it was removed, the business could breathe.
If you run a field service business and recognize any part of this story — the evening quote writing, the leads slipping through, the routes that could be tighter — book a free strategy call. We'll show you what this looks like against your actual data.
Running a field service business?
We'll show you exactly what an agent stack looks like against your data — quotes, routes, leads, reports. 20 minutes. No slide deck.
Book a Free Call