What makes Perspicacity different from a normal website agency?

Most agencies hand you a pretty website and walk away. We build a revenue system that works for you — the site, the AI receptionist, the follow-up, the CRM, and the reporting all designed together around one thing: turning more of your leads into booked, paid work you can actually measure.

What voice engine powers the AI receptionist?

The demo plays locally rendered neural speech samples; it is not a live production call. For a client line, we test locally run speech and language models on the actual call tasks, then separately validate the carrier path, latency, consent, transfers, logging, and CRM/calendar integrations before launch.

Can the AI receptionist provide medical, legal, or financial advice?

No — and that's deliberate. In high-trust fields the assistant is locked to intake, scheduling, triage, policy answers, and handoff. Real medical, legal, and financial judgment always stays with your licensed people. You get the speed without taking on the liability.

What does the delivery system standardize?

Everything that usually gets improvised and then forgotten — intake, page architecture, service packaging, schema, automation maps, QA gates, and launch runbooks. So your project starts from a complete, proven blueprint instead of someone figuring it out as they go.

Do clients need to buy every service at once?

Never. You start with the smallest engagement that can prove a revenue path, then add voice, GEO, paid acquisition, reactivation, or deeper automation only once the first one is working. You never pay for capability before you're ready to use it.

All packages

Perspicacity Package

Voice Operations Layer

An AI phone system designed to cover routine calls, offer appointment booking, and route follow-up when your team is busy or closed, with a tested fallback for exceptions.

Best for: Home services, clinics, property teams, appointment-heavy businesses

Request a fit call Compare all packages

Fit

Who this package is for.

Perfect fit

Businesses where the phone is how most customers make contact: home service companies (HVAC, plumbing, roofing), medical and wellness clinics, property management teams, and anyone running an appointment-based business. If you're missing calls after hours or your team spends too much time on phone tag, this is the fix.

Best for: Home services, clinics, property teams, appointment-heavy businesses

Terms you’ll see on this page — explained simply

Self-hosted voice: Speech and language models can run on servers we control when the components pass the required call tests. That can avoid a metered AI voice API, while the public phone number and call minutes still depend on a carrier.
Inbound vs outbound: Inbound means the AI picks up when customers call you. Outbound means the AI calls your existing leads or past customers — for appointment reminders, follow-ups, or re-engagement campaigns.
Consent and escalation rules: Before the AI goes live, we define and document: (1) what the AI is allowed to say and do, (2) exactly which situations it must immediately transfer to a human, and (3) how call recordings and transcripts are stored and disclosed to callers.
STT / TTS: Speech-to-text (STT) converts what the caller says into text the AI can process. Text-to-speech (TTS) converts the AI's response back into natural-sounding speech. Both run on our own infrastructure.
Webhook integration: A connection between the phone system and your other software (calendar, CRM, email) that fires automatically when something happens — like logging a new appointment or sending a confirmation text.

Scope

What’s included, explained simply.

Everything below is explicit. Nothing is implied or assumed to be included unless it appears in this list.

What you get — in plain terms

AI inference runs on servers we control, with the phone carrier and number provider disclosed in the scope. Carrier minutes, numbers, hosting, and support remain real costs and are priced transparently.
The AI can work inbound (answering calls as they come in) or outbound (calling old leads to re-engage them), depending on what your business needs. We configure and test the right approach for your situation.
Every script the AI uses is reviewed and approved by you before it goes live. We also define exactly when the AI hands the call to a human — for emergencies, sensitive situations, or any question it can't confidently answer.
The AI connects to your calendar and booking system, so it can check real availability and book appointments directly — no double-booking, no back-and-forth.
Where consent and integrations allow, calls are logged and transcribed for a regular sampled QA review. Failed or missing records are surfaced for investigation rather than counted as completed calls.

Full inclusions checklist

Self-hosted voice architecture
Inbound receptionist or outbound reactivation agent
Consent, escalation, and transfer rules
CRM/calendar/webhook integration
Transcript review and call quality QA

Boundaries

What’s not included — and why.

Knowing what’s excluded is just as important as knowing what’s in. If something you need isn’t on this list, it can be added as an expansion module or a separate engagement — ask during a fit conversation.

What’s not included — and the reason

We don't clone anyone's voice without explicit, documented consent from that person. This is both an ethical line and a legal one. The AI uses a professional neural voice persona, not an imitation of a real employee.
The AI doesn't give medical, legal, financial, or other professional advice. It can answer basic service and scheduling questions, but anything that requires professional judgment is immediately transferred to a qualified human.
Phone carrier costs (including minutes and numbers) and any separately scoped compute usage are tracked transparently. Locally run inference does not require a metered model API, but hosting and support still cost money.

Not included

Unconsented voice cloning
Regulated advice
Carrier and model usage fees

Items above require a separate engagement or expansion module. Ask about any specific capability during a fit conversation.

Timeline

How long it takes, and what happens when.

Timeline in plain terms

Three to six weeks from kickoff to live calls. The first two weeks involve designing the call flow, writing and approving scripts, and setting up the technical infrastructure. Weeks three and four are build and integration. The final phase is a full test run with real call scenarios before we go live.

Timeline and proof gates

Estimated timeline: 3-6 weeks
Infrastructure: $200-$900+/mo typical
Infrastructure detail: Model gateway, STT, neural TTS, SIP/telephony, call logs, transcript storage, and QA monitoring.

These are the conditions verified before we call the system ready:

Approved scriptsHuman transferUsage budgetTranscript and consent logs

Pricing

How the price works.

Pricing shown here is a planning range. We confirm fit and any needed scoping before setting final scope, timeline, fees, and deliverables in a signed statement of work.

How the price works — in plain terms

Setup is $4,500 to $12,000 paid once. The range depends on how complex your call flows are, how many lines or locations need coverage, and whether you need inbound, outbound, or both. The monthly fee of $750 to $2,500 covers system maintenance, script updates, QA review, and performance monitoring. Carrier minutes, phone numbers, and any separately scoped compute usage are itemized in the proposal.

Pricing and timeline

$4,500-$12,000$750-$2,500/mo + usage

Setup covers strategy, design, build, integration, QA, and launch. Monthly fees cover hosting support, optimization, reporting, and maintenance. Usage fees (telephony, SMS, model calls) are separately scoped where volume varies.

Timeline: 3-6 weeks
Infrastructure: $200-$900+/mo typical

Bundle math

Inbound reception, scheduling, and follow-up in one layer, with fewer duplicated tools than separate answering, booking, and reactivation stacks.

Floor is $4,500 (vs the $3,500 a-la-carte voice-reception setup) because the package bundles scheduling and lead reactivation on top of inbound reception — additional scope, not a different price for the same thing.

Expansion path

Where this package leads.

Expand into multilingual reception, after-hours dispatch, reactivation campaigns, and QA dashboards.

Ready to scope this package?

A short fit conversation helps us understand your bottleneck and decide whether a paid Diagnostic Sprint or a direct build scope is appropriate. Any paid work starts only after you approve its scope and price.

Request a fit call Compare all packages