Easy to start.
Room to grow. Costs that work.
Selected launch pricing: $39 Starter, $99 Pro, $249 Growth. Sell helpful voice and chat support for businesses, with simple self-service setup. Compatibility with a customer’s AI assistant is an optional convenience, not the headline.
These selected launch plans replace the earlier $149/$399 pricing hypothesis and are now shown on the customer-facing pricing page and in product facts. All package features below are production targets; checkout and live service are not connected.
A small-business entry point
| Included | Starter | Pro | Growth | Enterprise |
|---|---|---|---|---|
| Monthly price | $39 | $99 | $249 | Custom quote |
| Annual, paid upfront | $390 $32.50/mo equivalent | $990 $82.50/mo equivalent | $2,490 $207.50/mo equivalent | Contracted |
| Annual saving | $78 / 16.7% | $198 / 16.7% | $498 / 16.7% | Negotiated |
| Web voice minutes/month | 90 | 250 | 650 | Committed pool |
| Text-only AI conversations/month | 200 | 500 | 1,250 | Committed pool |
| Recording retention | 30 days | 90 days | 180 days | Contracted retention |
| Websites / team members | 1 / 2 | 1 / 5 | 3 / 10 | Contracted |
| Simultaneous calls | 1 | 3 | 5 | Reserved capacity |
| Extra voice minute | $0.30 | $0.25 | $0.20 | Costed quote |
| Extra text conversation | $0.10 | $0.08 | $0.06 | Costed quote |
All allowances reset monthly, including annual subscriptions. No rollover or unlimited promise. Annual savings apply to the subscription only; extra usage remains monthly. Website allowances and simultaneous calls share the same workspace usage pool.
- Every paid plan: the same core voice quality, website chat, multilingual conversations, website/FAQ knowledge, WordPress or script installation, lead capture, email/contact handoff, transcripts, summaries, recordings, basic analytics and spend controls. Security, deletion and export controls belong in every plan.
- Pro adds: booking/CRM workflows as they become available, more team access, richer reporting and configurable action policies. Removing Vozubi branding is reasonable here. Do not sell unbuilt integrations as included today.
- Growth adds: multiple websites, consolidated reporting, team permissions and faster business-hours support. It is for operational needs and higher usage, not automatically the right plan for every business with 500 conversations.
- Enterprise: custom capacity, SSO, contractual service levels, procurement and retention requirements. Internally, consider a $999+ monthly commitment when the scope warrants it, then cost the actual contract. No generic promise of unlimited usage or dedicated staff.
Positioning: “24/7 voice and chat support for your website. Easy to set up. Clear on what happened.” Business owners use the dashboard directly. Optional external-assistant access follows the same permissions and quotas; it does not justify a separate premium.
Free testing with a known acquisition cost
Offer a one-time Free allowance with 10 voice minutes, 25 text conversations and 7-day recording retention. No deadline to use the allowance; it does not renew. At the base assumptions below, a fully used Free allowance costs about $1.20 in AI usage; the stress case is $2.20, before setup and other operations. One Free allowance per verified business, rate limits and a budget for free usage keep this bounded. Once used, AI support pauses until upgrade; the business setup stays saved.
One hundred fully used Free allowances cost about $120 in base AI usage. If ten become paying customers, that is $12 of free usage per acquired customer, before advertising or human help. This is an acquisition expense, not free infrastructure.
What customers actually pay
A headline price is not enough. Change the workload and see all three plans, including overages. The lowest price is highlighted in words; the tool does not assume that the highest tier is the best fit.
| Plan | Base / month | Extra usage | Customer bill | Our modeled variable cost | Contribution before shared costs |
|---|
What if every included unit is used?
This second table uses each plan’s entire allowance, regardless of the workload entered above. It avoids relying on unused allowances to make the economics work. Cost and billing controls still apply.
| Plan | Monthly revenue | Variable cost | Contribution | Contribution margin |
|---|
For the default 30% voice / four-minute-call mix: 100 contacts cost $48 on Starter; 300 cost $121 on Starter or $126.50 on Pro; 500 cost $186.50 on Pro. These are monthly subscriptions with approved overages. At 300 contacts, Starter remains slightly cheaper; Pro’s additional workflow/team features must earn the upgrade.
Why these limits are defensible
OpenAI lists GPT-Live-1 at $0.05 per session-minute, billed per second, with backend models/tools charged separately. Our $0.07/minute base and $0.12 stress are planning assumptions for voice plus associated reasoning, retrieval, summaries and transcription where required—not a fixed all-inclusive supplier quote. Text estimates are $0.02/conversation base and $0.04 stress. Real language, context, reply counts and workflow calls must be measured. OpenAI pricing.
The model adds a $2/$4/$8 per-workspace monthly allowance for variable operations and assumes US domestic card processing of 2.9% + $0.30, plus Stripe Billing’s 0.7%. Annual subscriptions amortize their one payment across 12 months; an overage invoice adds its own $0.30. International payments, conversion, tax tooling, disputes and custom contracts may cost more. Stripe Payments · Stripe Billing.
This is contribution, not profit. At full allowance, the base model leaves approximately 58–59% on annual subscriptions before shared hosting, human support, sales/marketing, refunds, tax and founder pay. The higher-cost scenario lowers annual contribution to roughly 31–32%. These are scenario results, not guarantees; costs above the stress case remain possible.
With an assumed $100/month shared infrastructure bill, four fully used monthly Starter subscriptions cover that bill and their modeled variable costs. If we also spend $10/customer/month of human support, it takes six. Ten such Starter customers leave only about $81/month after those costs, before acquisition and founder pay. The $100 figure is a launch budget hypothesis, not a fixed hosting price at every scale.
Do not bundle unlimited personal setup help into $39. Keep ordinary self-service onboarding included; test a separately priced, clearly scoped assisted setup service if customers need it. Audit cost per tenant and per action, and require an economic review before enabling costly tools or provider options.
Recordings should build trust
Include recordings in Free and every paid plan, with 7 / 30 / 90 / 180-day retention. Capture eligible calls when recording is enabled and the required notice/permissions are in place. Show recording status and any capture failure, rather than promising every call is always saved. Owners should be able to disable recording, delete it and export it before expiry.
At an assumed aggregate 64 kbps, 1,000 minutes produce roughly 0.48 GB. Retaining three months of that monthly volume is about 1.44 GB, or roughly $0.022/month in raw R2 Standard storage before free allowances, requests and processing. Storage alone is not a reason to reserve recordings for expensive plans. Reliable capture, transcription, access control and playback are the engineering costs. Cloudflare R2 pricing.
Basic transcripts and summaries should follow an explicit deletion policy too; propose the same retention as recordings initially so owners understand what will remain. Longer retention is convenience and governance value. Avoid a separate recording-per-minute fee at launch; include expected recording operations in unit-cost measurements.
Overages without unpleasant surprises
- Meter connected web calls by the second. Do not round each call to a whole minute. Failed connection attempts are not customer usage. Idle and maximum-duration controls prevent abandoned sessions from running indefinitely.
- Define a text-only conversation. One visitor thread in a rolling 24-hour window, up to 40 AI replies. Count only when a useful AI reply is delivered. At the reply limit, explain the limit and offer handoff; do not silently create paid blocks. A thread that becomes a call is billed by voice minutes instead of a text-conversation unit. Transcripts and in-call information cards are not extra conversations. Measure the cost of these mixed sessions before launch.
- Extra spending is opt-in. Default to the included allowance; let owners enable overages and set a dollar ceiling. Notify at 80% and 100%. When a budget is exhausted, stop new AI sessions and offer a contact form or human route; gracefully finish active calls within reserved budget.
- Show upgrade comparisons. Estimate the current bill versus the next plan. Never assume an upgrade saves money, and do not change plans without the owner choosing it.
- Separate telephone-network service. These rates cover calls inside the website. Phone numbers, PSTN/SIP transport, SMS and premium third-party actions need a separately costed offer if added later.
Competitive enough to test, not a cheapest-price promise
Reception’s monthly $29/$79/$199 plans offer shared credits: web minutes use 0.5 credit and text messages use 0.1. A $79 plan’s 275 credits can cover 550 web minutes or 2,750 text messages, not both at those maxima. It also includes booking, summaries and transcripts. At our 300-contact example, assuming four billable text messages per text conversation, 264 credits fit its $79 plan. Vozubi’s $121–$126.50 modeled bill is higher. We should be candid about that; the offer must win customers through the experience and useful business workflows, not a false claim of price leadership. Reception billing.
A $39 entry is still much easier to try than $149. Core voice/chat quality, a clear owner activity record, dependable handoff and straightforward installation matter more than inventing an AI-builder positioning. Validate willingness to pay with direct business owners as well as agencies.
Launch gate: benchmark at least 100 realistic calls and 100 chat sessions across short/long conversations, intended languages and tools. Measure typical and expensive sessions, mixed text-to-voice usage, failures and support time. Then run a small paid pilot cohort. Keep the selected launch prices fixed during that cohort; increase allowances later if measured economics support it. Avoid lifetime discounts and unlimited offers.