How We Turned Repeatable Client Work Into a Productized AI Service

For the past year, the work we do for field-service clients has followed a pattern. An electrician calls with a lead intake problem. A landscaper needs invoicing automated. An HVAC company's CRM doesn't talk to their scheduling tool. We map their workflows, build the connections, train someone on their team, and move on.

Every engagement looked different on the surface. Under the hood, they were the same four phases: understand what's broken, spec the fix, build it, support it. We kept reinventing the wrapper around the same core work.

So we stopped doing that.

The Operations Engine

We launched the Operations Engine as a formal productized service for small service businesses: electrical, HVAC, plumbing, landscaping, construction, ag distribution, light manufacturing. The target client has 3 to 50 people, a few disconnected tools (CRM, scheduling, accounting), and an owner spending too many nights catching up on paperwork.

The four services are the same things we've always built:

• Quote and lead pipeline. A new lead hits the site or inbox, gets scored, a first response is drafted in the owner's voice, and it lands in the CRM. Three minutes after it arrives, not three days.
• Office automations. Scheduling, invoicing, follow-ups, route optimization, job sheets, weekly reports, wired together with n8n across the 400+ apps it supports.
• Job history memory. Every quote, job, and customer note becomes searchable. The system knows who you served two years ago, what you charged, and whether they're due for a return visit.
• Operations audit. Before any build, we map the business: top 10 manual workflows scored by impact and effort, the three that move the needle, a real ROI estimate.

Why Four Fixed Phases

The free-form "AI consulting" engagement has a trust problem. The client doesn't know what they're buying. We don't know what we're selling until we're already in it. Scope creeps in both directions.

Fixing that meant writing down the phases and making each one a named deliverable with a defined output:

1. Audit (Week 1). 90-minute discovery call, then five days of async work. The output is a written Operations Map: the top 10 workflow leaks, the top 3 to fix, and a dollar-and-hours cost estimate for each. The client keeps this document whether they hire us for the next phase or not. That guarantee does a lot of selling on its own.

2. Build Plan (Week 2). For each of the top 3 fixes, we write a technical spec and an ROI projection. The client sees exactly what gets built, what it costs, and what it returns before committing.

3. Install (Weeks 3-6). We build the engine, integrate it with existing tools, and run weekly demos so the client sees it working before the next milestone is billed. One person on their team gets trained to extend it without us.

4. Run (Ongoing, optional). Monthly tune-up call, one new workflow added per month, SLA-backed support.

The Pricing Structure

Three tiers:

• Operations Audit: $1,500 flat, one week. The standalone deliverable. Useful even if nothing else gets built.
• Done-with-You Install: $7,500 flat, four weeks. Audit plus build plan plus a live engine with 5-8 codified workflows and 2 live automations.
• Engine + Retainer: $12,500 install plus $1,500/month. Adds a custom AI assistant, Slack and email integration, and ongoing monthly additions.

The install tier is structured to close in a month. Most of our clients don't want a six-month engagement with a moving finish line. A defined four-week window with weekly demos makes the decision easier.

What We Built to Support It

The Operations Engine landing page has its own audit request form with a Zod-validated API route that writes directly to Airtable and fires a Resend email with the inquiry details. PostHog tracks audit form submissions as a named event so we can see what traffic actually converts.

We've had broken lead capture before. The new setup bypasses n8n entirely for form handling. Each API route writes to Airtable and sends the email inline, no external dependencies in the hot path.

The page also includes an explicit "this isn't for you if" section. We don't work with clients who want a giant SaaS rollout or who just want to "explore AI." We only take on operators who know what's broken and want it fixed.

That filter saves everyone time.