Built to be heard.
A prospect's name, title, and company go in. Ten seconds later, a bilingual voicemail comes out — drafted by an LLM, spoken in a cloned voice, ready to play mid-call.
Hear it, not just read about it
Same prospect, same pipeline, both languages — no password needed for these two.
Marie Lavoie, Director of Digital Services at a Canadian bank — drafted and spoken by the exact pipeline below, one call per language.
Why bilingual, specifically
Federally regulated Canadian institutions, and increasingly any business operating in Quebec under Bill 96, are on the hook to serve customers in both official languages. Today that usually costs a bilingual hire, a translation pass, or a second studio session per script. This tool collapses that into one extra API call: the same anglophone rep can leave a fluent French voicemail for a Quebec account, same day, without adding headcount. That's the actual argument for building this — not a language toggle, a coverage model.
What changes for a rep
The ICP here isn't every prospect — it's any account where speed-to-follow-up and language coverage both matter: federally regulated institutions, anything Quebec-facing, anywhere a same-day bilingual voicemail beats a callback that never happens. The day-to-day shift for a rep isn't more calls, it's fewer follow-ups that fall through the cracks between them — a name goes into a form while the last call is still fresh, and a personalized voicemail is in the prospect's inbox before the next one starts.
Enter the password to open the tool
This unlocks the actual generator — type in a company, prospect, and title, and it drafts, synthesizes, and plays the voicemail back in your browser.
An interactive preview of the CSV, Slack, and WhatsApp rows below — WhatsApp specifically pending Meta template approval, the others just unbuilt.
This started as a rehearsed demo — two languages, two scripts, hand-approved, zero live generation. It became something else the moment the question changed from "can we generate a good voicemail?" to "can we generate one for whoever's on the other end of this call, right now?"
You type a name into a form while you're still talking. Before you've finished the sentence, the room hears a voicemail — in your voice — for a company you just heard of ninety seconds ago.
That's the pitch this tool exists to make in front of a live audience. Everything below is how it actually works.
One pipeline, one branch point
What each stage actually does
Draft
An LLM call takes company, prospect, and title and writes a ~70-word script sized for voicemail, in the right language, ending with the callback number spelled out digit by digit so it's easy to catch by ear.
Synthesize
The script is rendered through a Professional Voice Clone, on the specific model that clone was actually fine-tuned for — not the newest model, the correct one.
Deliver
The mp3 either plays back immediately in the browser, or gets emailed as an attachment — the second path always shows exactly who, what, and how big before it will send anything. Cold commercial email in Canada falls under CASL, the same regulatory lens the roadmap applies to live dialing below — this demo's gate gets you explicit consent to send, not full CASL compliance (sender ID, unsubscribe) yet.
Scale
The same three stages run over a CSV of prospects instead of one form submission — every script drafted and shown before a single credit is spent on audio, every recipient shown before a single email goes out. Two languages, four pipeline stages, two confirmation gates, zero messages ever sent without an explicit yes.
Two ways to run it, on purpose
The tool actually ships as two separate paths, because they carry two different kinds of risk.
The demo pipeline
Rehearsed
Two languages, hand-written, human-approved before anyone ever heard them. Zero live generation at demo time — a bad line gets caught by you, in rehearsal, not live in front of an audience.
The live pipeline
On-demand
LLM-drafted, per prospect, in seconds — unreviewed by a human before it's spoken. The tradeoff is made up for downstream: nothing leaves the machine to a real person without an explicit confirmation first.
Dead ends, caught before they cost anything
Every delivery-channel claim below sounded plausible and turned out to not hold up — caught by checking the actual docs and terms, not by trusting the marketing page.
Doesn't exist. The Twilio integration is live phone calls only — no file delivery over any channel.
Against the rules. LinkedIn's API has no third-party DM endpoint; automating it via the UI violates the User Agreement.
Wrong platform, wrong shape. OpenAI's iMessage plugin is a macOS desktop feature with no API — nothing a server can call.
Real, but gated. Twilio can do it, but only behind a Meta-approved message template — an approvals project, not a code change.
Same three stages, more doors out
The draft → synthesize core doesn't change. What changes is who triggers it and where the result goes.
CSV bulk upload
Already works from the command line, one row at a time or a whole list — a real prospect CSV, two review gates, nothing sent unseen.
Slack as a second front door
Not a replacement for the live-call form — too slow for that — but a way for teammates to queue up voicemails asynchronously, ahead of a call.
WhatsApp delivery
Technically buildable on Twilio's API today. Blocked on a Meta-approved outreach template — paperwork, not engineering.
Live outbound dialing
Having the clone actually place the call means trading this TTS pipeline for ElevenLabs' Conversational AI product — and running into robocalling law the moment a human hasn't opted in first.
iMessage delivery
OpenAI's ChatGPT plugin can send iMessages now, but it's a macOS desktop feature with no API — nothing a server can call. Dead end, not a queue.
Preview the roadmap → — an interactive walkthrough of the CSV, Slack, and WhatsApp rows above, pending Meta template approval for WhatsApp specifically.
Built to be heard, not just read
Same rep, same voice, same tool — pennies per prospect, not a studio session.