Voice AI Day.
By 5pm your team is shipping calls.

1 day, on-site Up to 8 attendees Pricing on request UK, US, Australia

Most voice AI evaluations end the same way: three vendor demos, an internal write-up, no decision. We run a workshop that ends with a working voice agent on a real telephony number, against your CRM, against your knowledge base, in your tone of voice. By 5pm the team has done the production engineering the demos skipped, and the build-or-buy question has answered itself.


A working voice agent. Yours. Owned, deployed, documented.

The agent is yours, the credentials sit in your environment, the code goes home with the team. We do not pitch a vendor and disappear; the production-relevant patterns we use on our own paid voice AI builds get installed in the team's hands directly.

Voice sales qualifier

Inbound or outbound. Qualifies against your ICP, books meetings into your reps' calendars, pushes notes to HubSpot or Salesforce.

Voice tier-one customer service

Reads your help centre, your release notes, your terms. Cites the source on every answer, escalates the rest with full context.

Voice receptionist

Always-on phone-answering agent that books meetings, takes messages, routes urgent calls to a human. Sounds like a person who works at your company.


Vendors run good demos. Production engineering is different.

Vapi, Retell, ElevenLabs, and the other voice AI vendors all run good demos. The demo is not the problem. The problem is the engineering between "the demo works" and "this runs against our actual CRM, our actual telephony, our actual knowledge base, with audit logging the compliance team will accept." We do that engineering on the day.

The team learns where each vendor breaks. By the end of the workshop you know whether voice AI is the right move for this use case, with evidence. Bottie is live in production at bottie.bot with paying users, on the same stack the workshop installs in your team's hands. Most consultancies that pitch voice AI have shipped none of it.


Hour by hour.

AM

Morning

Pick the use case. Wire the voice infrastructure (Vapi or LiveKit). Connect speech (Deepgram, ElevenLabs). Get the agent making test calls before lunch.

MID

Midday

Plug into your knowledge base for cited answers, into your CRM for note-taking and meeting booking. Test with realistic call flows the team brings in.

PM

Afternoon

Audit logging, escalation flows, fallback paths, voice tuning to your tone, the compliance overlay (TCPA-aware for US, Ofcom-aware for UK). Production polish.

END

End of day

Working agent on a real number, owned by your team. Architecture diagram, runbook, vendor-cost model at your projected volume. 30 days of email support included.


The shape of buyer who lands here.

01

VP Sales / Head of SDR evaluating voice qualification before committing to a six-figure vendor contract.

02

Head of Customer Experience or Customer Service Director where tier-one volume is the bottleneck.

03

Head of People or L&D exploring voice-first onboarding, induction training, voice compliance training.

04

CTO or Head of Engineering embedding voice into a product, wanting the team to learn the production patterns properly.


The full voice AI stack.

Voice infrastructure

Vapi LiveKit Twilio Voice

Speech

Deepgram (STT) ElevenLabs (TTS) Cartesia

Reasoning

Claude API OpenAI Function calling MCP

Integrations & compliance

HubSpot / Salesforce / Pipedrive Zendesk / Intercom Calendar APIs TCPA / Ofcom audit logging

Book Voice AI Day.

Discovery call first. We confirm the use case, agree the date, and send a one-page brief two weeks before so the workshop lands against your specifics.