Hi,
I’m looking for someone with expertise in building AI voice agents in Eleven Labs. Ideally also with experience within Azure and Deepgram. If you know anyone or this sounds like you could you please get in touch on [email protected]
Thanks
Fiona
1 Like
Hi Fiona, I build AI agents for businesses. I co-founded Watbot, a platform where more than 70,000 bots have been built since 2018; they answer customers from a knowledge base and pass tricky questions to a person. I would build your Eleven Labs voice agent the same way: script, knowledge base, speech recognition through Deepgram, and Azure for hosting and data if you need it. I suggest starting with a free prototype for one scenario, for example answering a call and handling common questions, so you can hear the result before committing. Let me know what the agent is for, and happy to continue by email.
Max
Hi @Fiona_Challis,
I build AI voice agents with ElevenLabs, and I’ve worked with Deepgram for transcription and Azure on the cloud side.
Happy to help. Emailing you now, but here is the short version: I run Pathfinder Automation Solutions, Make.com, n8n and Zapier Expert Top Rated on Upwork with 100% job success. Most of my voice work is agents that handle inbound calls, qualify, book into a calendar and write back to the CRM.
What is the agent doing, inbound, outbound, or both?
Taiwo
Email: [email protected]
Website: https://pathfinderautomationsolutions.com
Upwork: https://www.upwork.com/freelancers/pathfinderautomate
Book a call: Calendly
Hi Fiona, i just sent you an email 
.
Just in case you can book a call: HERE
We have built plenty of AI agents, both text and voice.
Thank you for replying to my post in the Make.com community. We’ve had around 30 responses, so I’m writing to everyone with more detail to help us shortlist.
Background
We originally built our voice agents in Azure, using an API connection to Deepgram, as we wanted to keep everything within the Azure and M365 stack. The quality hasn’t been where we need it. We’ve had pauses, unnatural responses, trouble with background noise and inaccurate speech-to-text, especially when capturing email addresses. We’ve now moved to a full build in ElevenLabs, potentially still using Deepgram, and need an experienced specialist to help us finish, tune and deploy it.
The use case: two voice agents
Agent 1: Inbound AI receptionist
- Answers calls on our clients’ existing phone numbers.
- Mostly handles new customers, so there’s no existing data to look up.
- Captures the caller’s name, email address and preferred callback number (which may differ from the number they’re calling from).
- Books an appointment and sends the calendar invite.
- Must work for callers on both mobiles and landlines. Voice agents keep mishearing email addresses, even with spell-back. Texting a link works for mobile callers, but we need a reliable solution for landline callers too.
Agent 2: Outbound qualification agent
- Calls leads we already hold in our system, to qualify them and warm them up.
- Confirms existing details, asks qualifying questions, books a follow-up where appropriate and writes the outcome back to the CRM.
What we’re looking for
- Proven ElevenLabs Conversational AI experience with live deployments.
- Telephony experience. You’ve deployed agents over standard phone lines (e.g. Twilio or SIP trunking), not just web or desktop widgets.
- Natural conversation, with low latency, no awkward pauses, interruption handling and resilience to background noise.
- Integration with calendars, CRM and M365 via Make.com or webhooks.
Timeline
We’d like to start as soon as possible.
Next step: a phone demo, please
Given the number of replies, please send us a phone number we can call to test a voice agent you’ve built. Voice quality over a phone line is very different from a web or desktop demo, so this is the most important part of the shortlisting. Please also include:
- A brief description of the agent and the stack behind it.
- How you’d reliably capture a new caller’s email address over the phone, including from a landline.
- Your availability to start.
- Your day rate, or a fixed-price estimate for both agents.
We’ll test every demo number we receive and come back to the shortlist by [date].
Thank you again for your interest.
Kind regards,
Fiona Challis
Hi Fiona, I’ve just emailed you about the ElevenLabs voice agents. We build custom voice agents with ElevenLabs, Deepgram for transcription and Azure for hosting, and I’m happy to share a plan and a fixed quote once I know what the agents need to handle.
Rimal, AIDEM LABS
Hi Fiona, thanks for the additional details. I’ve just sent you an email with the demo number and a bit more context on my experience. Looking forward to hearing your thoughts
Hi Fiona, thanks for the detailed follow-up. I’ve emailed you our answers on the stack, how we’d reliably capture email addresses from landline callers, and our availability to start.
Rather than a standalone number, we’d like to show you both agents live on a short call. We can have the agent ring your own mobile and landline during the session, so you hear exactly how it sounds over a real phone line and can test email capture yourself.
Hi Fiona, thanks for the detailed follow-up.
So you want two ElevenLabs agents, an inbound receptionist and an outbound qualifier, that sound natural on a real phone line and book straight into calendar and CRM.
How I’d approach it: ElevenLabs agents on Twilio numbers (your clients forward their existing lines), Make.com handling calendar, M365 and CRM steps via webhooks, with CRM writes done after the call so the caller never waits on them. For landline email capture I wouldn’t rely on spell-back alone: split the address into username and domain, check the domain live during the call, and never let the booking depend on the email.
I’ll email you full answers to your questions and a fixed quote. Three quick ones back:
1. Which CRM and calendar (Outlook/Bookings)?
2. Are your clients’ numbers on Teams Phone or a regular carrier?
3. Roughly how many calls a month per client?
[email protected] or DM me here.
Tana, Tana Web Systems
Hi Fiona. Do you have a separate, small webhook/calendar/CRM integration task for around US$50? We would use AI-assisted development with human review. We cannot claim prior live ElevenLabs deployments. Your existing accounts would cover provider costs. Could you confirm the scope and whether payment to a contributor in Guatemala is possible?
Hi Fiona,
Thanks for the detailed brief. Here’s what you asked for.
Demo number: (805) 519-8885
This is my own receptionist. It answers, finds out what the caller needs, takes their name and callback number, and books a real slot on my calendar during the call. Try it from a landline too. It’s a US number, so it may be an international call for you.
The stack: Retell runs the agent with an ElevenLabs voice, Twilio carries the phone line, and booking goes into Cal.com and Google Calendar. To be straight with you, my live agents are on Retell, not on ElevenLabs’ own agent platform. The phone line, booking and webhook work is the same, and I’m happy to build yours in ElevenLabs.
Getting the email right, landlines included:
- During the call, check whether the caller is on a mobile or a landline. Twilio can tell you that for a fraction of a cent.
- Mobile: text them a link and they type the email themselves.
- Landline: ask for it in two parts, the bit before the @ and then the domain. Most people use Gmail, Outlook or a few others, so the domain is usually a pick from a short list. If it’s a business domain, check it really takes email before reading it back. Read it back letter by letter, “B as in Bravo.”
- The booking never waits on the email. The callback number is just digits, and agents get those right. If the email still isn’t certain, the slot is held and the record is flagged in the CRM so your team confirms it on the callback.
I’ll send my start date and rate to your email.
Cindy Garcia
Advanced AI Automation
advancedaiautomation.com/call
Hi Fiona, this sounds relevant to the work I do. I have hands-on experience with AI voice agents, Vapi, Make, APIs and webhooks, including connecting voice workflows with CRMs and other business systems.
I haven’t worked directly with every tool in the stack, but I’m comfortable working with voice APIs and integrating them into automation workflows.
I’ll send you an email with some examples of my work.
Hi Fiona, we build ElevenLabs voice agents at Octacer, both inbound and outbound, connected to telephony and the client’s CRM so calls actually turn into bookings or tickets. Is this agent going to answer inbound calls, or make outbound ones like follow-ups?
Hi, I’m Mikael – I build Make.com scenarios for voice AI (ElevenLabs, Azure, Deepgram), telephony, and CRM integration.
AI voice agents in Eleven Labs + Azure/Deepgram.
I build Make scenarios that connect ElevenLabs conversational AI, Azure Speech (STT/TTS), Deepgram streaming, and Twilio/Vonage voice. Last build: inbound call → ElevenLabs agent → CRM log + Slack alert, 200 calls/day, <500ms latency.
$500 fixed per scenario or $800/mo retainer (voice minutes extra). Includes test number, call logs dashboard, and fallback handling.
Mikael
Hi Fiona,
I saw your post about ElevenLabs voice agents. We work on AI automation and conversational workflows at LeadFlow AI.
Is this opportunity still open? We’d be interested in learning more about the requirements.
— Ihor & Vladimir
Hi Fiona — are you still assigning paid work? I operate VYTREL with AI-assisted delivery and founder oversight. My demonstrated work is an owned Make-to-Supabase inspection and deployed API/reconciliation system; I do not have ElevenLabs/Azure/Deepgram production-client experience to claim. If you have a separate, contained API/webhook reliability or data-mapping task around the voice agent, I could review a redacted brief before confirming fit and a fixed quote. Please share the input/output, failure example, budget and deadline. Paid API usage stays on your accounts with your approval. We would agree scope, deposit and acceptance before work. This is an inquiry, not acceptance of the full build.
For building an AI voice agent using Eleven Labs, I would set up a trigger that activates when a user sends a voice input. The workflow would include steps for processing the audio through Eleven Labs for voice synthesis, integrating with Azure for any additional AI functionalities, and utilizing Deepgram for speech recognition. Error handling would involve logging any issues during processing and providing fallback responses. The final deliverable would be a detailed JSON workflow file, along with documentation for setup and usage. If you need clarification on specific features or integrations, please let me know.
Gilbert, GL Tools (Vienna, written communication)