Apex Climate Solutions — Cost page

Apex would cost about $59 to $121 a month to run (estimate) and, in the base case, save about $1,035 a month after paying its own running cost, or $12,420 a year (modeled). The cost follows request volume, not team size: 330 to 550 requests a month, and AI phone answering is over half the cost. At the same call volume, Apex on Jobber's entry tier costs about the same as Jobber's own AI receptionist and review tools ($307 to $369 against $332 to $384, keeping Jobber's review add-on for review requests); the case for Apex is the features it adds.

Apex Climate Solutions is a simulated company, built as a portfolio demonstration. Most figures here are estimates or modeled; vendor list prices and the measured Make credits and token counts are labeled as such. No shop has used Apex, and prices were read on October 25 to 27, 2026, and are not kept current.

Savings estimate

Base case (modeled): Apex frees about 30 staff hours a month, worth about $1,094 in staff pay including benefits (about 21 office hours at $34.92 an hour and 9 technician hours at $40.95). Minus the $59 a month it costs to run Apex itself (broken down under the Running cost section), that is about $1,035 a month saved, or $12,420 a year. Apex does not remove people from the process. It shortens their part, and each saving is the time a task takes today minus the time a person still spends on it.

All three columns are stated estimates. The review numbers assume the full design: new reviews pulled in automatically and approved replies posted back. The demo builds the drafting and approval steps; the connections to Google and Instagram need a real business's accounts, so they are designed, not built. The example shop has five technicians doing three jobs a day over 22 working days, so 330 requests a month, one per job. Staff time is priced at US national wages from the Bureau of Labor Statistics (office $24.44 an hour, technician $28.66) times about 1.43 for benefits, giving $34.92 and $40.95 an hour.

The arithmetic, base case

1. Office triage: 330 requests × 3.5 min ÷ 60 = 19.3 hours.
2. Review replies: 10 a month × 8 min ÷ 60 = 1.3 hours. Office total 20.6 hours × $34.92 = $719.
3. Technician questions: 5 technicians × 1 a day × 22 days × 5 min ÷ 60 = 9.2 hours × $40.95 = $375.
4. Value: $719 + $375 = $1,094 a month.
5. Net: $1,094 − $59 running cost = $1,035 a month, $12,420 a year.

The catch. Saved time becomes saved money only if the shop spends less on staff, for example by skipping a part-time hire or paying less overtime. If only half the freed hours lower the payroll, the base case drops to about $5,900 a year. Even so, Apex pays for itself if it saves about 18 seconds of office time per request. Running cost barely moves the result; the time-saved estimates and whether hours become money decide it.

Vendor pricing

For a 5-technician shop choosing one system to run the business, the established field-service vendors cost $299 to $2,500 a month. That covers the everyday tools such a shop needs (customer records, job tickets, scheduling and technician assignment, invoicing), plus AI call answering, review requests and replies, and connections to its other tools. Housecall Pro's and ServiceTitan's AI phone add-ons, and ServiceTitan's review module, have no published price and aren't included. Prices are list, annual billing, shown per month, read from each vendor's own pages on October 25, 2026, unless marked. Sales tax, card-processing fees and Jobber's limited-time first-year discount are left out.

Apex alone is not a full field-service system. It has no dispatch calendar, invoicing or payments, so the fair comparison is Apex running on top of an entry tier. Apex doesn't send review requests, so at the same call volume the fair comparison keeps Jobber's $99 review add-on on the Apex route. The two routes then cost about the same: the Apex route is $15 to $25 a month less (estimate). Sending Apex's job records into Jobber is design only, so this compares running costs only.

Jobber's AI receptionist is not a flat fee below its top plan. On Connect it is a $29 add-on covering 30 conversations, then $0.79 for each one, so Jobber's bill rises with calls too. Its Marketing Suite, for review requests and replies, is $99. Above about 184 AI conversations a month, Jobber Plus becomes cheaper at a flat $399 (5 logins) or $428 (6) with unlimited AI answering (Jobber help center, read Oct 26, 2026).

What Apex adds to Jobber. Vendor research on September 12 to 13, 2026 found that Jobber does not offer WhatsApp intake, a graded urgency rating (its AI receptionist escalates on keywords only), technician answers drawn from the shop's own manuals, truck-stock tracking or permit tracking. Apex covers each of these: WhatsApp intake, a 1 to 5 urgency rating, cited answers from the shop's manuals, a truck-stock lookup, and a watchdog that flags overdue jobs, low stock and permits stuck in pending. Apex can also stand in for Jobber's AI receptionist, which keeps the combined cost about level with Jobber's own AI route.

Running cost

Apex would cost about $59 to $121 a month (estimate) at 330 to 550 requests a month, and AI phone answering is over half of it. That is the example 5-technician shop at 3 to 5 jobs per technician a day, or $12 to $24 per technician; a larger team pays more only because more requests come in. The range assumes AI answering every call, Slack on its free plan and a hosted office screen.

This is what Apex costs to run. Setup, upkeep and any fee are left out. Each phone call costs about 27 to 40 cents in AI voice minutes (the measured $0.134 a minute, 2 to 3 minutes a call), while a WhatsApp request costs well under a cent in message fees, so steering customers to WhatsApp is the simplest way to lower the bill.

A measured demonstration month used more Make credits than the design estimate. The design estimate put a 100-request demonstration month at 84% of Make's 10,000-credit plan. Measuring a real demonstration month found 11,593 credits, 116% of the plan. Four fixes followed:

• The overdue-job watchdog now runs every 30 minutes inside working hours instead of around the clock.
• WhatsApp delivery receipts and early phone-call events now stop at the first step instead of running the whole flow.
• The technician question answerer merges its variable steps.
• Old test jobs and mock stock records were cleaned out.

After the fixes the same month projects to about 8,188 credits, 82% of the plan. That projection is the basis for the shop figures above (about 71 credits per request, plus about 1,100 a month for scheduled checks). It has not been measured over a full month.

Cost cut tested: prompt caching

Caching the fixed part of the urgency rater's instructions cuts its modeled monthly cost by 63% to 73%, so the 1-hour cache was adopted for the urgency rater and the junk filter. Prompt caching lets an AI model reuse the unchanging front of a prompt at a fraction of the normal input price. It pays only if calls arrive close enough together to find the cache still active.

What was measured on a test copy of the urgency rater, never the live one:

• The fixed instructions are 1,235 tokens, above the model's 512-token caching minimum.
• The first call wrote the cache. Every later call read all 1,235 tokens from it.
• A call more than 5 minutes after the last found the short cache expired and wrote it again.
• Prices read from Anthropic's pricing page on October 26, 2026, per million tokens: normal input $4, cache read $0.20, 5-minute cache write $5, 1-hour cache write $8.

The monthly figures are modeled from the measured token counts and random arrivals across a 12-hour day. The cache breaks even when at least 20.8% of calls find it active (5-minute) or 51.3% (1-hour). Below about 4 requests a day, the 1-hour cache costs more than it saves. On the junk filter the saving is about a cent a month.

Two setup problems would have stopped caching without any error, and both were caught before caching was adopted.

• The original Make request template could not carry the cache marker, so a new template was built that stores the fixed instructions as a marked block.
• Automatic caching marks the last block of the prompt, which here is the customer's message and changes every time. It would never have read the cache, so the marker sits on the fixed instructions instead.

The request extraction step cannot cache: its fixed prompt is below that model's 4,096-token minimum.

What is and isn't counted

The savings figure counts what is built, plus review pulling and posting, which is designed and needs a real business's accounts; other design-only items and benefits that are hard to price are left out.

A benefit not counted: calls a busy shop would otherwise miss. Call-tracking vendors estimate that home-service shops miss 14% to 27% of their calls, and some of those callers book with someone else. If Apex's AI answers every call, the calls this shop would otherwise miss could be worth roughly $9,000 to $22,000 a year. The savings figure leaves this out for three reasons: Jobber's own AI receptionist would catch the same calls, so it is not an Apex advantage; the miss rates come from companies that sell call handling; and it is extra income, not saved cost.

Other limits

• Setup, upkeep and any fee are excluded.
• Every time-saved input is a stated estimate. Timing 20 requests in a real shop, before and after, would replace them.
• Some AI and phone usage behind the running cost is assumed, not measured: the request extraction tokens and real call lengths.
• Wages are US figures to match the US software prices. A shop in a lower-wage market frees the same hours but banks fewer dollars.

What changes at real volume. Jobber's own route stops rising once Plus, at a flat $399 to $428 with unlimited AI answering, is the cheaper choice; Apex's cost keeps rising with messages. At 30 to 40 messages a day, every day, the running cost rises to about $173 to $273 a month (estimate) with paid Slack. Run on Jobber Connect with Jobber's review add-on, that is $421 to $521, more than Jobber Plus at $399 to $428, so at high volume Apex costs more than Jobber's own route and its case rests on what it adds.

Sources

Vendor and price pages were read October 25 to 27, 2026, unless noted.

• Vendors: Jobber pricing · Jobber Receptionist terms (Oct 26) · Jobber and Zapier · Housecall Pro pricing · Housecall Pro CSR AI · ServiceTitan pricing (no published prices)
• ServiceTitan estimates, unverified third-party: OneCrew · Projul
• Wages and benefits: BLS, dispatchers · BLS, HVAC mechanics and installers · BLS, employer compensation costs, June 2026
• Missed calls and repair value: Invoca (vendor) · CallRail report, January 2025 (vendor) · Fixr, HVAC repair cost
• Running cost: Claude API pricing (Oct 26) · Gemini API pricing · WhatsApp Business Platform pricing · Render pricing · Neon pricing · Slack pricing · Make prices from third-party guides (Activepieces), unverified
• Project records: the Retell per-minute rate measured on the Apex phone agent; Make operations measured over a demonstration month; the prompt-caching test runs; the project charter's list of what is built, design only and out of scope. The full return-on-investment calculation and the full pricing comparison are companion documents behind this page.