The dispatch board is the hardest hidden math in the shop. At 11:45 a.m. on a Tuesday, a high-value call lands — Bel Air homeowner, blower motor down, equipment over 14 years old, the kind of call that historically converts to a $14K replacement quote — and the dispatcher has to decide in 90 seconds whether to send Jose (38% MPR, sells, but is 45 minutes out on a recall), pull the rookie off a low-value drain (eats the dispatch fee on the drain but burns a quick training rep), or hold the call for a 2:30 slot when Marco wraps the install (but Marco's hands are gold on linesets and pulling him off the install costs $4K on the day's gross margin). That decision is not a calendar app. That decision is the difference between $620 dispatch yield and $707 dispatch yield, which on a 7-truck day works out to ~$345K of annual margin contribution at typical residential service economics. ServiceTitan Dispatch Pro is built to make that decision automatically every 10 minutes using predicted job revenue × tech historical close rate × travel × capacity. Sera Systems takes a profit-aware approach optimizing revenue per tech. FieldEdge auto-routes against capacity and skill tags. ResponsiBid drafts the bid before the tech opens the tablet. This lesson is the role-by-role tour of what the AI is actually deciding on the board and on the truck — and what the dispatcher and the tech still own.

What ServiceTitan Dispatch Pro Is Actually Optimizing For

The reflex when an operator first sees Dispatch Pro's screen is to assume it is doing magic. It is not. Dispatch Pro is doing fast regression on four variables every 10 minutes across every open call and every available tech. The four variables are: predicted job revenue (what this call type has historically generated at this shop, adjusted for system age, equipment type, customer history, and time-of-day patterns), tech historical close rate (this tech's documented close on this type of call), travel cost (driving time from current location to the next stop, weighted against shop fuel cost and lost billable time), and capacity (the tech's remaining hours on the day, the install crew's stacked dependencies, the recall windows). The output is a ranked recommendation: send this tech to this call now, push this call to the 2 p.m. slot, hold this call for tomorrow because all qualified techs are revenue-disadvantaged on it today.

The re-evaluation cadence is the differentiator. Old-school dispatch logic re-runs at most a few times a day — when calls land, when techs finish, when emergencies hit. Dispatch Pro re-runs the entire board every 10 minutes. That cadence matters because the shop's reality changes every 10 minutes: a tech finishes early and pulls up; a parts run extends a stop; a homeowner cancels; an emergency lands. The 10-minute re-evaluation captures every reshuffle opportunity and pushes the dispatcher an updated recommendation. The dispatcher decides whether to override.

The published 2026 outcome across deploying shops is 12-18% dispatch yield lift in 60 days when paired with human override discipline. Override discipline means the dispatcher overrides 5-15% of Dispatch Pro's recommendations with documented reasons (comp plan, install crew, recall risk, customer request) and the model handles the rest. The dispatcher's role evolves from board-mover to exception-handler. The shop captures the lift; the dispatcher captures the higher-leverage work; the model captures the override patterns and tunes against them over the following months. Cost: Dispatch Pro is a ServiceTitan add-on around $200-$400/month per shop. The payback is in the first three weeks.

Sera Systems and the Profit-Aware Flip

Sera Systems' 2026 dispatch AI takes a different bet. Where Dispatch Pro optimizes the predicted-revenue × close × travel × capacity formula, Sera optimizes against a single target metric: revenue per tech. The bet underneath is that call-count optimization tolerates low-margin calls (the AI fills the day with billable activity but the day's margin is thin), whereas revenue-per-tech optimization explicitly de-prioritizes filler work when a higher-revenue alternative is reachable.

In practice the difference shows up on slow days. At 1:30 p.m. on a slow Tuesday with three available techs and four open calls — one Bel Air replacement quote, two service tune-ups, one warranty callback — Dispatch Pro's call-count instinct is to fill the slots: each tech runs a call, the day is full. Sera's profit-aware instinct may hold one tech for a higher-revenue call expected to land in the afternoon based on call-volume patterns, dispatch only the Bel Air to the highest-close tech, and route the two tune-ups to a single tech for routing efficiency. The fewer-calls-but-higher-revenue outcome is the Sera bet. The shop's quarterly RPT (revenue per truck) is the metric that proves the bet right or wrong over a 60-day window.

Neither model is universally better. A high-volume residential service shop with thin margins per call may benefit more from Dispatch Pro's call-count optimization. A replacement-heavy shop with significant revenue spread across call types may benefit more from Sera's revenue-per-tech focus. The decision tree, in trades English: shops where average ticket variance is high and replacement attach is a big lever, lean Sera. Shops with stable ticket and high call volume, lean Dispatch Pro. The bake-off lives in Level 4 Chapter 2; the role-level introduction is here.

FieldEdge Auto-Routing and the Windshield-Time Question

FieldEdge's smart routing AI operates on a third axis: travel time and clustering. The bet is that on a service-heavy day with 5-7 trucks across a metro service area, the difference between a thoughtful route and a chaotic route is 60-90 minutes of windshield time per truck per day. At a $80-$120 fully-loaded hourly cost per tech, 5 trucks × 75 minutes/day × 250 days = ~$200K of annual recoverable labor cost — before counting the additional billable calls a 75-minute time savings enables.

FieldEdge's auto-routing clusters geographically, weights against skill tags (a tech who is fast on heat pumps gets the heat pump cluster), respects time-window commitments (the 10 a.m. confirmed slot must run at 10 a.m. regardless of clustering optimization), and re-routes when the day's mix changes. The dispatcher's override authority sits in two places: stacked install crews (do not break the install crew for service routing efficiency) and recall returns (the recall goes back to the originating tech, period). FieldEdge's 2026 outcome data shows 15-20% windshield time reduction at shops with disciplined override protocols.

The shop-by-shop pick between Dispatch Pro, Sera, and FieldEdge often follows FSM platform: ServiceTitan shops use Dispatch Pro; Sera-platform shops use Sera's dispatch AI natively; FieldEdge shops use FieldEdge auto-routing. The 59% in-software AI preference in ServiceTitan's 2026 State of AI in the Trades report shows up most strongly in dispatch — bolt-on dispatch AI is unusual because the integration depth required (every booking, every tech location, every job status update in real time) is hard to maintain across a vendor boundary.

ResponsiBid and AI Quoting at the Bid Step

The dispatch decision feeds the bid decision. Once Dispatch Pro routes Jose to the Bel Air call, Jose needs a bid in the customer's hand inside the same visit at a 50-60% close-rate-target shop. The bid-drafting time used to be the bottleneck — a Comfort Advisor took 25-45 minutes per home assembling a good/better/best three-option proposal with financing math, rebate math, photo annotations, and warranty language. ResponsiBid's 2026 AI quoting expansion collapses that drafting time to 8-12 minutes.

The mechanic: the tech (or the Comfort Advisor on the ride-along) pastes the equipment scope, the homeowner's stated goals (lower bill, longer warranty, quieter system), the financing tier (Wisetack / GreenSky / Synchrony approval level if pre-pulled), the rebate context (state and utility incentives applicable to this zip), and the photo set. ResponsiBid generates the three-option narrative, computes the financing payment math at each tier, pre-formats the warranty language, and annotates the photos with the AI tagging engine — corrosion, refrigerant stain, clearance issue, panel deficiency. The tech reviews, edits the personal-touch sections, and presents — 30 minutes saved per proposal, with documented $200-$600 average-ticket lift per closed deal because the proposal looks more thorough and the good/better/best logic lifts mid-tier and high-tier selection.

ResponsiBid combined with ServiceTitan estimate templates and AI narrative drafting is the 2026 bid-generation stack. Cost: ResponsiBid runs $400-$800/month per shop in 2026. Payback inside the first month at typical service-volume economics. The discipline that protects the lift: the 30-second verify on every AI-drafted proposal (SEER/AFUE numbers, financing payment math, warranty term, rebate amount within current state-utility table) before the customer sees it. Without the verify, fabricated rebate amounts and miscomputed payments ship to the kitchen table and the close-rate lift evaporates under trust damage.

What the Tech Does With AI on the Truck

The tech's day has five places where AI compresses time or lifts ticket. The lesson here is what the tech is doing with AI — the deeper workflow design lives in Level 2 Chapter 4. Right now the goal is to name the touch points so the tech understands what the apprentice is doing and what they sign off on.

Voice Notes Into Structured Tickets

The tech finishes a service call and dictates voice notes into the tablet. The AI structures the notes into the format the service manager actually reads: equipment make and model, age, condition, recommended next step, customer mood, follow-up window. The tech speaks for 60 seconds; the AI produces a structured ticket in 15 seconds. Before AI, the tech either typed for 12 minutes or skipped the documentation and the dispatcher chased them with a phone call at 4 p.m. The structured ticket feeds the dispatcher, the service manager, and the next visit. The tech's verify: read the AI output, fix the equipment make if it heard wrong, fix the customer name if it hallucinated, send.

Repair-vs-Replace Prompts at the Customer

The tech opens a 14-year-old furnace, finds the heat exchanger compromised. The conversation with the homeowner is whether to repair (replace the heat exchanger, $1,400) or replace the system ($8,900 mid-tier, $14,200 premium). The AI takes equipment age + repair cost + customer goals + utility rebate + financing payment and outputs a 3-option talk-track the tech reads off the tablet: repair (here is what it costs, here is what it does not solve), partial (replace the furnace, keep the coil, $5,400), replace (here are the three options, the rebate, the financing payment). The pricing model AI presentation discipline — good/better/best — produces 52-58% mid-tier close at shops that present this way (Comfort Advisor training history confirms the pattern). The tech is reading the AI output; the homeowner is hearing a coherent three-option conversation; the close rate lifts because the conversation is grounded rather than improvised.

Photo Annotation and Tagging

The tech takes 8-12 photos per service call. The AI tags each photo automatically: rust, corrosion, refrigerant stain, clearance issue, panel deficiency, drain belly. The annotated photos drop into the proposal automatically with explanatory captions. Average ticket lift on photo-annotated proposals: $180-$320 in published 2026 trades data — partly because the homeowner sees what the tech sees, partly because the photo-grounded proposal looks more thorough than a text-only quote. The tech's verify: confirm the AI tagged the right element (a refrigerant stain is not a water stain; an old breaker is not a deficient one without the actual measurement), edit captions for accuracy, send.

Membership Pitch Tailored by AI

The tech wraps the service call. The AI generates a 30-second membership pitch tailored to the customer's specific equipment age, system condition, prior repair history, and average annual repair cost. The pitch reads naturally because it is built on what just happened, not on a generic script. MPR (Membership Penetration Rate) lifts from 22% baseline to 35-50% target at shops with this discipline; top-quartile Nexstar shops run 60%+ MPR. At a typical $14-$22/month membership × 1,200 customers × 25% incremental penetration = $50K-$80K of annual recurring revenue per shop. The tech's verify: confirm the equipment age and repair history pulled correctly into the pitch before saying it.

Self-Dispatch and Route Editing on the Tablet

Sera and Housecall Pro's 2026 pilots shift more of the dispatch role to the tech. The AI proposes the day's route; the tech edits on the tablet (skip the lunch stop in Reseda, push the 2 p.m. call to 3:00 because the parts run hit traffic); the dispatcher becomes the exception handler rather than the operator. The pilot works in small teams with geographically tight territory and high-trust techs. It breaks in larger teams where training calls need centralized routing, callbacks need to return to the originating tech, and install crew coordination requires central control. The Level 2 Chapter 4 lesson on self-dispatch covers the operational depth; the role-level mention here flags that the tech's day is increasingly co-authored with the AI rather than dictated by the dispatcher.

What the Dispatcher Still Owns When the AI Reshuffles Every 10 Minutes

The dispatcher's reflex when Dispatch Pro lands is "the AI is taking my job." The honest answer is the role evolves rather than vanishes — the dispatcher becomes the exception handler, the override authority, and the comp-plan-vs-yield interpreter.

Five categories of override sit permanently in human hands. Comp plan dynamics. Marco is on a PIP and a low ticket finishes him; Dispatch Pro doesn't know that. Jose is on a streak and the bonus tier kicks in at one more close; Dispatch Pro doesn't see the comp curve. The dispatcher overrides the routing to manage humans, not just metrics. Install crew gross margin. Pulling Marco off installs to handle a service call costs $4K on the day's lineset margin; Dispatch Pro's predicted-revenue formula doesn't capture that opportunity cost. The dispatcher protects the install crew. Recall risk by tech. The recall on the Marin Park job needs to return to the originating tech (Marco) because Marco's hands on the original install reduce re-callback probability; sending the rookie to a recall doubles the recall risk. The dispatcher routes recalls by origin, not by capacity. Customer request specificity. "Send Jose. I want Jose." Dispatch Pro doesn't see customer-tech history; the dispatcher does and honors the request. The unseen variable. The dispatcher knows the weather, the traffic, the rookie's mood, the recent argument with the supplier, the construction project blocking a street. None of those are in the model.

The override discipline is logged. Every override gets a one-line reason in Dispatch Pro's system; the dispatcher's weekly review at the Friday standup pulls the override patterns and feeds them back to the model and the comp plan. Shops that log overrides systematically see Dispatch Pro improve over months — the model learns the shop's hidden variables and reduces the override rate from 15% in month 1 to 5-8% in month 6. Shops that override without logging stay at 15% forever because the model never sees why.

The Metrics the Dispatcher and the Tech Watch Together

Three numbers move when AI lands on the dispatch board and the truck. The dispatcher and the tech watch them together because the numbers are coupled — a routing decision feeds a tech outcome feeds the next routing decision.

Dispatch yield (revenue per dispatched call). Baseline ~$620 at typical residential service. Target +12-18% in 60 days with Dispatch Pro / Sera / FieldEdge plus override discipline. At a 7-truck × 6-call/truck day × 250 days × $87 per-call lift, the annual margin contribution lands at ~$345K. This is the headline dispatch-row number.

RPT (revenue per truck per day). 2026 baseline: shop baseline $1,600-$2,400/day for residential service trucks per Built on Tenth, Level CFO, and MarginPlug data. Top-quartile target $2,800-$3,500/day for service trucks, $2,500-$4,000/day for replacement trucks, $5K+/day for commercial replacement. RPT is the tech-row scorecard; dispatch yield × calls per truck per day = RPT. Average residential service ticket lands at $450-$600 median, $800-$1,500 high-mix.

MPR (Membership Penetration Rate). Baseline 22%; target 35-50%; top-quartile 60%+. The AI-generated membership pitch tailored to equipment age and repair history is the lever. Each MPR point at a 1,200-customer shop × $18/month average × 12 months = $2,600 of annual recurring revenue per point — and the recurring revenue compounds because members convert at higher rates on the next replacement.

Key Takeaways

• ServiceTitan Dispatch Pro re-evaluates the board every 10 minutes using predicted job revenue × tech historical close × travel × capacity. Published 2026 outcome: 12-18% dispatch yield lift in 60 days with override discipline. Cost $200-$400/mo.
• Sera Systems optimizes revenue per tech rather than call-count. Profit-aware scheduling favors fewer, higher-revenue calls on slow days. Better fit for replacement-heavy shops with high ticket variance.
• FieldEdge auto-routing cuts windshield time 15-20% on service-heavy days. The dispatcher overrides stacked install crews and recall returns; the model handles the rest.
• ResponsiBid plus ServiceTitan estimate templates plus AI narrative drafting cuts proposal time from 25-45 min to 8-12 min and lifts average ticket $200-$600 per closed deal. Cost $400-$800/mo. Payback inside month one.
• The tech's five AI touch points: voice-to-structured-ticket notes, repair-vs-replace 3-option talk-track, photo annotation and tagging, AI-generated membership pitch, self-dispatch route editing on the tablet.
• Dispatcher override authority lives permanently in five places: comp plan dynamics, install crew gross margin, recall risk by originating tech, customer request specificity, and unseen variables (weather, traffic, mood, supplier dynamics).
• Override logging is the discipline that makes Dispatch Pro improve. Override rate drops from 15% month 1 to 5-8% month 6 at shops that log; stays at 15% forever at shops that don't.
• Three numbers move: dispatch yield (baseline $620, +12-18% in 60 days, ~$345K margin contribution), RPT (baseline $1,600-$2,400/day, top-quartile target $2,800-$3,500/day), MPR (baseline 22%, target 35-50%, top-quartile 60%+).
• The dispatcher role evolves to exception-handler and override-interpreter. The tech's day is increasingly co-authored with the AI — voice notes, prompts, photo annotations, membership pitch — but the journeyman sign-off discipline still applies to everything customer-facing.