AI-Generated Replace-vs-Repair Proposals for HVAC

Why the replace-vs-repair conversation is the install business

The HVAC install business is decided in a forty-five minute conversation at a kitchen table. The system is failing. The customer is stressed, partially informed, and weighing $389 against $14,400 against doing nothing for another summer.

In that conversation, the contractor who arrives with a complete picture — equipment age, service history, charge readings, rebate eligibility, financing pre-approved at three terms, a tiered good/better/best proposal sized to the home's load — closes at materially higher rates and tickets than the contractor who promises to "send a quote tomorrow." AI compresses time to assemble that picture from 90 minutes at the office to ten minutes on the customer's couch.

What the tablet workflow looks like in 2026

The tech opens the FSM mobile app and triggers the proposal flow. Behind the scenes, AI assembles four streams of data and renders them as a tiered proposal in under three minutes.

Stream 1 — equipment context

Pulled from the FSM install record. Make, model, age, refrigerant type, prior service tickets, warranty status, any open technician recommendations. If the FSM record is incomplete, AI reads the rating plate photo the tech captured and fills in the gaps from the manufacturer's database.

Stream 2 — load-calc assist

The AI reads square footage, window count, insulation grade, and zone configuration from the homeowner's input plus the tech's walk-through observations and produces a defensible Manual J sizing recommendation. The certified installer reviews and signs; the AI does not autonomously size.

Stream 3 — rebate and incentive stacking

Federal IRA tax credits (25C and 25D), state rebates, local utility rebates, and manufacturer instant rebates are stacked automatically against the customer's zip code and the proposed equipment. The savings line on the proposal is the real net cost after stacked incentives — not a number that erodes in the financing conversation.

Stream 4 — financing pre-approval

A soft credit pull via Wisetack, Synchrony, or GreenSky returns pre-approval at three terms (typically 0% for 18 months, low APR for 60 months, low APR for 120 months) in under 60 seconds. The customer sees monthly payment numbers attached to each tier of the proposal, not just sticker prices.

The output is a three-option proposal — typically a like-for-like replacement, a mid-tier with efficiency and modest comfort upgrades, and a premium with the inverter heat pump, smart thermostat, and 12-year extended warranty. Each tier shows the gross price, stacked rebates, net cost, and monthly payment.

Dynamic estimating of this shape lifts average install ticket 6–9% on consistent deployments — published outcome data from ServiceTitan Pricebook AI, Sera Estimates, and Housecall Pro's proposal builder converges on similar numbers.

Tools an HVAC operator should evaluate

ServiceTitan Pricebook AI

The deepest integration for ServiceTitan shops. Strong on tiered proposal layout and rebate-stacking. Outdated SKUs and labor rates produce outdated proposals — pricebook hygiene matters.

Sera Estimates

Strong on dynamic pricing tied to capacity. Best fit over 10 trucks.

Housecall Pro proposal builder

Solid for 2–8 truck shops. Easier than ServiceTitan, less granular on tiered logic.

Coolfront / Profit Rhino

Flat-rate pricebook overlays for shops without a strong native FSM pricebook.

Claude or ChatGPT for proposal narrative

The paragraph explaining why this customer should choose this tier is meaningfully better when drafted by frontier models. Most shops use this as last-mile polish.

Manual J load-calc assist — the new pattern

Through 2024 most HVAC shops ran Manual J only on permitted installs where the AHJ required it, and even there the calc was a 90-minute office task in Wrightsoft. AI economics shift the calc onto every replacement proposal.

The pattern: the tech captures dimensions, window count and orientation, insulation grade (best-guess), zone configuration, and current performance complaints. The AI produces a defensible sizing recommendation, a heat-loss/heat-gain calc per zone, and a homeowner-readable explanation. The certified installer reviews, adjusts, and signs.

This shaves 35–50 minutes off every install consult and produces a proposal narrative that outperforms "your current system is old" as a sales rationale.

Implementation — getting the proposal flow live

Step 1 — clean the pricebook

The single biggest predictor of proposal-flow success is pricebook hygiene. SKUs missing from the FSM, labor rates that have not been updated since 2023, accessories priced inconsistently across tiers — all of these produce bad proposals. Budget 30–60 hours of pricebook cleanup before the AI layer goes live.

Step 2 — map the tier templates

Define the three tiers as configurable templates. Good = direct replacement, baseline efficiency. Better = mid-tier efficiency, smart thermostat, 10-year warranty. Best = highest efficiency inverter heat pump or variable-speed system, smart thermostat, 12-year warranty, indoor air quality add-on. The templates are configuration, not per-tech freelancing.

Step 3 — integrate rebate stacking

Federal, state, utility, and manufacturer rebates by zip code. This is the layer that vendors handle differently — some maintain the rebate database internally, some pull from external services like EnergyStar's REBATES feed. Verify your tool's rebate coverage in your service area before committing.

Step 4 — financing integration

Wisetack, Synchrony, or GreenSky pre-approval flows wired to the proposal. Confirm the soft-credit-pull UX is one-tap on the tablet, not a separate browser session.

Step 5 — pilot with two senior techs

Two weeks. Senior techs only. They run the proposal flow on every install consult, the office manager reviews every proposal against the pre-AI baseline, and the prompts get tuned.

Step 6 — full rollout

Once override rate is under 10% and senior tech feedback is positive, every tech doing install consults gets the flow. Training is one 90-minute session plus shadowing.

Pitfalls

Outdated pricebook

The most common failure. AI optimizes against bad input and produces a polished bad proposal. Fix the pricebook first.

Treating the AI sizing as autonomous

Manual J recommendations from AI are an assist, not a finished calc. The certified installer signs. Skipping the human review produces sizing errors on edge cases and creates real liability.

Single-option proposals

Shops that present a single price miss the 18–24% ticket lift from tiered proposals. The discipline is to always present three tiers, even when the tech is confident the customer will pick the middle one.

Rebate misattribution

AI form-completion of rebate applications must attach the correct AHRI matchup certificate; the wrong cert produces a denied rebate weeks later and an angry customer. Verify the matchup logic during pilot.

Failing to update the financing offers

0% for 18 months expires; APRs change quarterly. The proposal flow must read live financing terms from the financing partner, not cached values.

FAQ

Will AI proposals replace the senior estimator?

No. They compress the estimator's per-job time from 90 minutes to 10 minutes and let one senior estimator effectively cover three to four times the deal flow. Most shops redeploy estimator capacity to higher-end commercial bids rather than reducing headcount.

Are the load calcs defensible to the AHJ?

Yes, when the AI assist is paired with the certified installer's review and signature. Some AHJs require Wrightsoft or equivalent output specifically; check your local AHJ before submitting.

How does the financing soft-pull affect the customer's credit?

Soft pulls do not affect credit scores. The customer sees pre-approval at terms before any hard inquiry. The hard pull happens only on accepted financing.

What is the realistic lift on average install ticket?

6–9% with consistent deployment. Higher lift is possible but typically reflects shops moving from no-tiered-proposals at all to tiered proposals on every install. For the full dollar math, see the HVAC AI ROI breakdown.

What happens when the customer wants a quote over the phone?

The AI receptionist captures the situation and books an install consult. Phone quotes are not a workflow the AI tries to complete — the conversion data on phone quotes is materially worse than in-home tiered proposals.

How does this interact with the AI receptionist?

The receptionist captures system type, age, and current symptoms. That data flows to the proposal engine before the tech arrives, which means the tech walks in with 80% of the equipment context already assembled. For the receptionist workflow, see the AI receptionist for HVAC guide.

Can the proposal flow handle multi-system homes?

Yes. The flow generates per-system proposals (e.g., first-floor mini-split + basement boiler + attic furnace) and consolidates them into one customer-facing document with combined financing. Confirm the multi-system logic with sample proposals during pilot.

For the broader operator context, see AI for HVAC contractors: the 2026 operator playbook and the AI enablement hub.

Ready to deploy a proposal flow on your trucks? Start at AI for HVAC contractors, or book a 30-minute consult and we will audit your pricebook, your current install close rate, and your average ticket against the benchmarks in this guide.

SOURCES

Cited and consulted.

1. 01ACCA Manual J Residential Load Calculationacca.org · accessed May 8, 2026
2. 02ServiceTitan Pricebook AI — Implementation Guideservicetitan.com · accessed May 8, 2026
3. 03Contracting Business — Replacement Sales and Proposalscontractingbusiness.com · accessed May 8, 2026
4. 04HVACR Business — Estimating and Saleshvacrbusiness.com · accessed May 8, 2026