Every trade has its own dictionary. HVAC has TXV, EER, MERV, AFUE, SEER2, HSPF, MBH, ASHRAE 62.2, and ACCA Manual J. Plumbing has DFU, GPM, IPC 802.6, backflow, AAV, ABS, PEX-A vs. PEX-B, and trap-arm-length-by-fixture. Electrical has GFCI, AFCI, AIC, NEMA, MCA, voltage drop, and NEC 250.122. Roofing has shingle squares, ridge cap, drip edge, ice-and-water shield, ASTM D3462, and the soffit ventilation ratio. The 2026 trades AI conversation adds another forty words you will hear at every vendor demo, every Nexstar Super Meeting, every ServiceTitan Pantheon panel, and every PE-platform quarterly review. This lesson is the trades-honest glossary — every word defined in shop language, with a CSR / dispatcher / tech / advisor / owner usage example for each. Bookmark it. Reference it. By the time you finish Level 2, half of these will be muscle memory. By the time you finish Level 5, you will be defining them for someone else.

AI Machinery — The Eight Words That Describe the Tool

Agent. An AI that takes actions in your shop's systems, not just produces text. Avoca answers a call and books the appointment in ServiceTitan — that booking action is what makes it an agent. Hatch sends a nurture text and transfers a warm lead — that is agent work. Dispatch Pro reshuffles the board — agent. ChatGPT, out of the box, is a chatbot, not an agent, because it produces text and stops. CSR usage: "The Avoca agent booked us four after-hours calls last night — I need to verify the slots before the techs roll." Owner usage: "Every agent we deploy needs an action-audit workflow before it goes live."

Hallucination. The AI confidently producing fabricated content. The 12 lb R-454B charge that should be 9.3 lb. The Section 25C credit cap that is wrong by $3,000. The fabricated NEC reference. Hallucination is not a bug; it is the predictable output of statistical text generation when training data is sparse. Tech usage: "I almost ordered the part the AI gave me — caught the hallucination on the manufacturer site." Dispatcher usage: "The AI summary said this was a return visit on Marco's territory — that was a hallucination, the customer is new."

Prompt. Your instruction to the AI. The five parts: role, context, task, format, constraint. CSR usage: "I built a prompt for the price-shopper rebuttal and now I save 4 minutes per call." Comfort Advisor usage: "My follow-up text prompt has the homeowner's wife's concern about warranty as the context."

System prompt. The AI's standing orders that apply to every conversation until changed. The vendor's actual moat — months of trades-specific tuning live here. Owner usage: "I asked Avoca what their system prompt does that we couldn't write ourselves — three things I would not have thought of."

Token. The unit AI counts and is billed on — a piece of a word. Marketing manager usage: "Our Hatch monthly invoice is calculated against the token usage of every text drafted."

Model. The specific AI you are renting (GPT-5, Claude Opus 4.6, Gemini 2.5, Llama 4, Mistral Large). Owner usage: "Rilla swapped models last quarter and our transcription accuracy improved 6%."

Temperature. The dial that controls how random/creative the AI's output is. Low for warranty, financing, and code language. Higher for brainstorming. Marketing manager usage: "I bumped the temperature up for the email subject-line drafts and got better hooks."

RAG (Retrieval Augmented Generation). The AI fetches the relevant chunk from your data (ServiceTitan history, manuals, transcripts) and feeds it as context. Most serious 2026 trades AI does this under the hood. Owner usage: "Titan Intelligence does RAG on our ticket history — that's why it knows our customers."

The Phone and CSR Words

Booking %. Calls booked / calls received. The single most important phone-row number. Industry median for an under-trained CSR floor is around 65%. Target with AI assistance is 80-85%. Top-quartile shops run 90%+. Each booking point ≈ $9K-$14K annual revenue at a typical 7-truck shop. Owner usage: "Booking % is up to 82 since Avoca went live — that's the only number on my dashboard that matters this quarter."

CSR show rate. Booked calls that actually ran. Industry baseline ~84%, target with AI on-my-way confirmations 92%+. Dispatcher usage: "Show rate ticked up after we turned on the AI-generated on-my-way texts."

Missed-call %. Calls that rang and were not answered or were abandoned. Industry baseline ~22% — the number Avoca was built around. Target with AI receptionist deployment: under 5%. Owner usage: "We were at 22% missed and now we're at 4 — Avoca pays for itself in the first three weeks."

After-hours capture rate. The percentage of after-hours calls that result in a booked appointment. Pre-AI baseline 0-15%. With AI receptionist 80%+. Owner usage: "After-hours capture went from 11% to 78% — that's a new revenue source the shop never had."

ACW (After-Call Work). The minutes a CSR spends writing notes, tagging the customer record, and disposition-coding a call after the customer hangs up. AI call summaries cut this from 4 minutes to 15 seconds. CSR usage: "ACW dropped from 4 minutes to 15 seconds since the AI summaries went live — I can handle 30% more calls per shift."

RPL (Revenue Per Lead). Revenue produced divided by leads generated. The marketing-side conversion metric. Marketing manager usage: "RPL is up 18% since we layered AI bid management on top of GLSA."

Cost per booked call. Total marketing spend divided by number of booked appointments. The right way to measure marketing efficiency in 2026. Marketing manager usage: "Cost per booked call dropped from $156 to $97 after we deployed Hatch on stale leads."

The Dispatch and Truck Words

Dispatch yield. Revenue per dispatched call. The dispatcher's hidden scorecard. Lifts 12-18% in 60 days when AI dispatch optimization is paired with override discipline. Dispatcher usage: "Dispatch yield went from $620 to $740 per call after we let Dispatch Pro re-evaluate every 10 minutes."

RPT (Revenue Per Truck per day). The truck-level productivity metric. 2026 baselines per Built on Tenth, Level CFO, and MarginPlug: shop baseline $1,600-$2,400/day for residential service trucks, top-quartile target $2,800-$3,500/day; replacement trucks $2,500-$4,000/day top-quartile; commercial replacement $5K+/day. Owner usage: "Our service-truck RPT is at $2,100 — top quartile is $2,800-$3,500, so we have room."

Dispatch Pro. ServiceTitan's AI dispatch optimization tool that re-evaluates the board every 10 minutes using predicted job revenue × tech historical close × travel × capacity. Owner usage: "We turned Dispatch Pro on six weeks ago and our dispatcher overrides have stabilized at 12% — the model handles 88%."

On-my-way text. The AI-generated SMS sent to the customer when the tech is en route — includes photo, bio, ETA, and review nudge. Lifts show rate 4-8 points. Tech usage: "The on-my-way text now has my photo and bio — customers greet me by name when I get there."

Self-dispatch. The 2026 emerging pattern where the AI proposes a route and the tech edits it on the tablet, with the dispatcher becoming an exception handler. Used in Sera Systems and Housecall Pro pilots; works best with small teams and tight territories. Dispatcher usage: "Self-dispatch handles the morning route; I focus on afternoon reshuffles and recall escalations."

The Sales Advisor and Coaching Words

MPR (Membership Penetration Rate). Percentage of completed service calls that result in a membership sale. Industry median 22%, target 35-50%, top-quartile Nexstar shops at 60%+. Service manager usage: "MPR is up from 22% to 41% since we built the AI-coached membership pitch into every service call."

Financing close rate. Percentage of jobs $5K+ that close with financing. Wisetack-published benchmark 28-40%. Industry under-trained baseline 14%. Comfort Advisor usage: "Financing close was 14% — we ran the AI-drafted pivot prompts and we're at 31%."

Wisetack / GreenSky / Synchrony. The three primary trades financing platforms in 2026. Wisetack for fast service-trade approvals and 0% promo windows. GreenSky for high-ticket replacement with deeper credit bands and longer terms. Synchrony for branded card programs and revolving lines. Comfort Advisor usage: "Soft-pulled the homeowner through Wisetack at the door, approved at $18K, kitchen-table conversation re-anchored around the payment."

Soft-pull. A credit check that does not affect the homeowner's credit score, used to pre-qualify them for financing before the formal application. Triggered by AI workflows pre-visit or at the door. Comfort Advisor usage: "Soft-pull came back at 740 — the financing pivot is going to be easy."

RGA (Ride-Along Grade/Score). Rilla-generated coaching score from a Comfort Advisor's kitchen-table audio. Five moments evaluated: intro, system-condition narrative, repair-vs-replace pivot, options presentation, financing pivot. Service manager usage: "Jose's RGA on the financing pivot dropped 0.4 last week — that's the coaching topic for tomorrow's huddle."

Knockout reason. The customer's stated reason for not buying — "I want to think about it," "I need to talk to my spouse," "the other quote was $4K less," "I'll wait until summer." AI tags these from kitchen-table audio for coaching. Service manager usage: "Last week 38% of knockout reasons were 'thinking about it' — we're rewriting the close prompts to address it."

Good/Better/Best. The three-option pricing model AI assembles for every replacement quote — sized to the customer's stated goals, with upgrade math (efficiency delta, comfort delta, rebate delta, financing payment delta) pre-computed. Mid-tier closes 52-58% in shops that present this way. Comfort Advisor usage: "I let the AI build the good/better/best from the system specs — the homeowner picked better, payment fit their budget."

The RC&D and Recall Words

RC&D (Recall, Callback, Warranty). The three categories of go-back work. Industry baseline 4-7% recall; top-quartile under 2%. AI classifiers triage these correctly so the shop measures the real recall rate. Service manager usage: "RC&D triage shows our true recall rate is 5.4 — we thought it was 3.2 because half were getting tagged as warranty incorrectly."

Recall. The same customer, the same complaint, the system is still broken. Internal failure. The shop eats it. Top-quartile under 2%. Service manager usage: "Three recalls this month on Jose's territory — pulling the tickets for root-cause."

Callback. The tech missed something on the original visit and the customer called back about it. Internal failure but different from recall — different remedy, different coaching topic. Service manager usage: "Callback rate dropped 3 points since we required AI job-notes on every ticket."

Warranty. The manufacturer part failed; the shop covers labor under warranty allowance. External failure. Different SLA tier (5 business days) than recall (24 hours) or callback (48 hours). Owner usage: "Warranty exception handling — when the manufacturer denies — escalates to me for the customer-recovery email."

Warranty exception handling. The workflow when the manufacturer denies a warranty claim. AI drafts apology + remedy script; owner signs off on shop-eats vs. partial-credit decision. Owner usage: "Warranty exception came in on the Goodman coil — manufacturer denied, AI drafted the customer recovery, I'm reviewing before it goes out."

The Marketing and Reputation Words

GLSA (Google Local Service Ads). Google's pay-per-lead local service ads. The dominant lead source for HVAC, plumbing, electrical, roofing in 2026. ROAS baseline 3-4x with mature optimization; AI bid management lifts 30-50%. Marketing manager usage: "GLSA ROAS is at 3.6 — we're trialing Ryze AI bid management for the 30-50% uplift."

GLSA AI bidding. Tools (Ryze AI and others) that layer ML on top of Google's bidding to optimize for booked calls rather than raw leads. Marketing manager usage: "GLSA AI bidding now disputes 18% more bad leads than we did manually."

Hatch. AI lead nurture platform. 2026 stale-lead reactivation lift: 30-45% across home-services deployments. Marketing manager usage: "Hatch revived $42K in dormant leads last quarter — 32% of them from the 90-day-old pile."

NiceJob / Podium AI Employee / Birdeye AI Employee / Yelp AI. Reputation management with AI-drafted review responses and sentiment-crisis flagging. 2026 cadence: 4-7 reviews per truck per month, 100% response within 48 hours. Marketing manager usage: "Birdeye AI Employee handles 90% of review responses; I review and post."

AEO (Answer Engine Optimization). The 2026 successor to SEO — optimizing content so AI engines (ChatGPT, Perplexity, Google AI Overviews) cite your shop when homeowners ask. 87% of HVAC and plumbing contractors are invisible in these answers (Plumbing & Mechanical 2026). Marketing manager usage: "Our AEO audit showed Perplexity cites our competitor for 'best HVAC company in [city]' — we're rewriting service pages this quarter."

CallRail Conversation Intelligence. The 2026 expanded CallRail product — call sentiment analysis, missed-opportunity flagging, lead-source attribution. Marketing manager usage: "CallRail flagged 8 missed-opportunity calls yesterday — three were CSR price-shock kills, two were after-hours leakage."

The Franchise and Platform Words

Authority Brands / Wrench Group / Apex Service Partners / Sila Services / Path Light Pro / Redwood Services / Leap Partners / ARS / Rescue Rooter. The 2026 PE-backed trades platform set — multi-shop / multi-brand consolidators operating across HVAC, plumbing, electrical, garage door. Multi-shop operator usage: "Our Authority Brands portfolio review is on the 15th — the AI dashboard is on slide 4."

One Hour Heating & Air / Benjamin Franklin / Mister Sparky / Aire Serv / Mr. Rooter. The major HVAC/plumbing/electrical franchise brands inside the Authority Brands / Neighborly portfolio. Franchisee usage: "I run two Mister Sparky locations — the franchisor-mandated tech stack is ServiceTitan plus the corporate-approved AI add-ons."

MAP (Marketing Assessment / Marketing Fund). The franchise marketing fund — typically 2-5% of revenue contributed to corporate for brand-level marketing. Affected by AI ROAS lifts. Franchisee usage: "MAP is 4% of revenue — corporate's GLSA AI bidding lifted the brand-level ROAS, which expanded our shared spend."

Technology Assessment Fee. A franchise fee for the corporate tech stack (often $200-$600/seat/month plus integration costs). Franchisee usage: "Tech assessment fee is $400/seat — and we get Avoca and Dispatch Pro included."

QBR (Quarterly Business Review). The franchisor's quarterly performance review with the franchisee — booking %, MPR, recall %, financing close, AI dashboard metrics. Franchisee usage: "QBR is next Tuesday — I'm prepping the AI-built one-page summary with the deltas vs. last quarter."

The Compliance and Regulatory Words

TCPA (Telephone Consumer Protection Act). The federal law governing automated calls and texts. AI voice agents and SMS bots have exposure if calling/texting wrong numbers or violating opt-outs. Owner usage: "We log every AI-initiated text in case TCPA inquiry comes in."

Two-party consent. The state-by-state requirement that all parties consent to recording a call. California, Pennsylvania, Florida, Illinois, Massachusetts among them. AI listening tools (CallRail, Avoca, Rilla) need disclosure handling. Owner usage: "Our two-party-consent disclosure plays at the top of every recorded call."

Reg Z (Truth in Lending Act). The federal financing disclosure rule. AI-drafted financing language has compliance exposure. Comfort Advisor usage: "Reg Z language has to match the Wisetack portal exactly — the AI draft gets a verify pass every time."

FCRA (Fair Credit Reporting Act). Federal law governing credit-pull disclosures and adverse-action notices. Comfort Advisor usage: "If the soft-pull declines, FCRA requires an adverse-action notice — the AI doesn't draft this; we use the lender's template."

EPA 608. The federal refrigerant-handling certification for HVAC techs. AI doesn't issue or verify EPA 608. Owner usage: "Every tech on the truck has EPA 608 — the AI doesn't substitute for that, ever."

FTC endorsement guidelines. The FTC rules governing reviews, endorsements, and (in 2026) AI-generated review responses and testimonials. Marketing manager usage: "FTC disclosure language sits in our AI review-response system prompt."

Reading the Vocabulary on Monday Morning

The owner's Monday standup, in 2026, is conducted in this vocabulary. "Booking % was 81 last week, missed-call % is at 4, RPT is at $2,180 for service and $3,100 for replacement, MPR is at 38, financing close is at 29, recall % is at 3.1, GLSA ROAS is 3.8 with AI bidding showing a 22% lift, Dispatch Pro overrides held at 14%, RGA average for Comfort Advisors is 4.2 out of 5, AEO audit shows we picked up two Perplexity citations, and the Authority Brands QBR is two weeks out." That entire paragraph is forty words from this lesson. The owner who can read it cold runs a different shop from the owner who needs each term explained. By Level 5, the operator running 25 locations will recite a version of this paragraph across every portfolio review and a PE board will see the discipline and the metric movement and approve the next add-on acquisition.

You do not need to memorize the glossary in one sitting. You need to know it exists, to reference it when a vendor uses a word you do not recognize, and to come back to it after every Level 2 lesson where the vocabulary gets used. By the end of Level 1, you should be conversant in the AI-machinery words and the phone-and-CSR words. By the end of Level 2, the dispatch and truck words. By the end of Level 3, the sales-advisor and RC&D words and the marketing words. Levels 4 and 5 polish the franchise/platform and compliance vocabularies. The progression is intentional. The glossary is your reference.

Key Takeaways

• The trades AI conversation in 2026 adds ~40 words to the trade vocabulary you already speak. The lesson groups them by where they live: AI machinery, phone/CSR, dispatch/truck, sales/coaching, RC&D, marketing/reputation, franchise/platform, compliance.
• Eight AI-machinery words you'll hear weekly: agent, hallucination, prompt, system prompt, token, model, temperature, RAG. Each has a CSR/dispatcher/tech/advisor/owner usage in this lesson.
• The phone-and-CSR vocabulary is the highest-leverage starting point: booking %, CSR show rate, missed-call %, after-hours capture, ACW, RPL, cost per booked call. Every shop tracks these or should.
• Dispatch/truck vocabulary: dispatch yield, RPT (with 2026 baselines and top-quartile targets), Dispatch Pro, on-my-way text, self-dispatch.
• Sales/coaching vocabulary: MPR, financing close rate, Wisetack/GreenSky/Synchrony, soft-pull, RGA, knockout reason, good/better/best.
• RC&D vocabulary: recall vs. callback vs. warranty (with SLA tiers — 24h / 48h / 5 business days), warranty exception handling.
• Marketing/reputation vocabulary: GLSA, GLSA AI bidding, Hatch, NiceJob / Podium AI Employee / Birdeye AI Employee / Yelp AI, AEO, CallRail Conversation Intelligence.
• Franchise/platform vocabulary: the 2026 PE platform set (Authority Brands, Wrench Group, Apex Service Partners, Sila Services, Path Light Pro, Redwood Services), the major franchise brands (One Hour, Benjamin Franklin, Mister Sparky, Aire Serv, Mr. Rooter), MAP, technology assessment fee, QBR.
• Compliance vocabulary: TCPA, two-party consent, Reg Z, FCRA, EPA 608, FTC endorsement guidelines. Each has direct AI exposure.
• Reference, not memorize. The glossary is your bookmark. By Level 2 the AI-machinery and phone-CSR words are muscle memory. By Level 5 you're defining them for your portfolio.