Hi everyone,
We’ve recently been working on an exciting project at Telecoms4Business, developing a custom AI-powered voice agent that integrates OpenAI with 3CX.
Our objective was simple: move beyond traditional IVR menus and create an intelligent virtual receptionist that can actually understand why someone is calling, assist with their enquiry and route them to the right person.
Rather than callers hearing:
“Press 1 for Sales, Press 2 for Support, Press 3 for Accounts…”
They can simply have a natural conversation with the AI.
What Have We Built?
We’ve developed an AI voice agent using OpenAI, integrated with 3CX and hosted on Amazon Web Services (AWS).
The system is designed to act as the first point of contact for incoming calls, understanding natural speech and determining the most appropriate next action.
1. Intelligent Call Routing
Instead of navigating complicated IVR menus, callers can simply say:
The AI understands the caller’s request and directs the call to the appropriate extension or department.
This provides a much more natural experience than traditional number-based menus.
2. AI-Powered Customer Support
One of the more interesting capabilities we’ve developed is allowing the AI to help resolve customer enquiries before transferring them to a member of staff.
For example, if someone calls regarding an internet connectivity issue, the AI can ask relevant troubleshooting questions.
Depending on the responses, it can guide the caller through potential solutions.
Where the issue cannot be resolved, the caller can be directed to the appropriate support team.
The idea isn’t to replace technical support but to reduce unnecessary transfers and allow engineers to concentrate on issues that genuinely require their attention.
3. Intelligent Sales Qualification
We’ve also developed conversational sales qualification.
Rather than immediately transferring every sales enquiry, the AI can ask a series of relevant questions to better understand the customer’s requirements.
For example, someone enquiring about a new business phone system might be asked:
The AI can adapt its follow-up questions based on the caller’s answers.
Once sufficient information has been gathered, the call can be transferred to the sales team.
This means the conversation has already progressed beyond the initial discovery stage before a salesperson becomes involved.
4. Natural Conversations Instead of Fixed Scripts
One of our main objectives was to avoid creating another rigid, scripted telephone system.
By using OpenAI, the agent can interpret different ways of asking the same question and respond conversationally.
For example:
“I need someone in technical.”
“My phones aren’t working.”
“Can you put me through to IT?”
These requests can be interpreted according to their meaning rather than requiring specific keywords.
The AI can also ask follow-up questions when the caller’s request isn’t clear.
How Is It Hosted?
Our solution uses:
Using AWS gives us flexibility over the hosting environment and allows us to develop additional functionality as the solution evolves.
Why Did We Build It?
We believe AI has the potential to significantly improve how businesses handle incoming telephone calls.
Traditional IVR systems are useful, but they often create unnecessary friction.
Customers don’t necessarily know which department they need. They simply want to explain their problem and speak to someone who can help.
Our aim was to create an AI receptionist that understands the customer’s requirements and takes the most appropriate action.
For businesses handling a mixture of customer service, technical support and sales calls, this could be particularly useful.
What’s Next?
We’re continuing to explore how conversational AI can improve the experience for businesses using 3CX.
Some of the areas we’re particularly interested in developing further include:
Interested in How We’ve Built It?
We’re keen to hear from other 3CX users and developers experimenting with AI-powered call handling.
Has anyone else developed a similar integration using OpenAI and 3CX?
It would be interesting to compare approaches, particularly around call transfers, conversational accuracy, latency and managing more complex customer enquiries.
For anyone interested in our work, the solution has been developed and configured by Telecoms4Business, a UK business telecommunications provider specialising in 3CX, VoIP, connectivity and managed IT.
Website: https://www.telecoms4business.co.uk/
Happy to discuss our approach and exchange ideas with other members of the 3CX community.
Thanks!
Telecoms4Business
We’ve recently been working on an exciting project at Telecoms4Business, developing a custom AI-powered voice agent that integrates OpenAI with 3CX.
Our objective was simple: move beyond traditional IVR menus and create an intelligent virtual receptionist that can actually understand why someone is calling, assist with their enquiry and route them to the right person.
Rather than callers hearing:
“Press 1 for Sales, Press 2 for Support, Press 3 for Accounts…”
They can simply have a natural conversation with the AI.
What Have We Built?
We’ve developed an AI voice agent using OpenAI, integrated with 3CX and hosted on Amazon Web Services (AWS).
The system is designed to act as the first point of contact for incoming calls, understanding natural speech and determining the most appropriate next action.
1. Intelligent Call Routing
Instead of navigating complicated IVR menus, callers can simply say:
- “Can I speak to extension 105?”
- “I need to speak to someone in accounts.”
- “I’m calling about a problem with my broadband.”
- “I’d like to discuss a new phone system.”
The AI understands the caller’s request and directs the call to the appropriate extension or department.
This provides a much more natural experience than traditional number-based menus.
2. AI-Powered Customer Support
One of the more interesting capabilities we’ve developed is allowing the AI to help resolve customer enquiries before transferring them to a member of staff.
For example, if someone calls regarding an internet connectivity issue, the AI can ask relevant troubleshooting questions.
Depending on the responses, it can guide the caller through potential solutions.
Where the issue cannot be resolved, the caller can be directed to the appropriate support team.
The idea isn’t to replace technical support but to reduce unnecessary transfers and allow engineers to concentrate on issues that genuinely require their attention.
3. Intelligent Sales Qualification
We’ve also developed conversational sales qualification.
Rather than immediately transferring every sales enquiry, the AI can ask a series of relevant questions to better understand the customer’s requirements.
For example, someone enquiring about a new business phone system might be asked:
- How many users require a phone?
- Are you currently using a traditional telephone system or VoIP?
- Do you have multiple office locations?
- Are you looking to retain your existing telephone numbers?
- Are there any particular features you’re looking for?
The AI can adapt its follow-up questions based on the caller’s answers.
Once sufficient information has been gathered, the call can be transferred to the sales team.
This means the conversation has already progressed beyond the initial discovery stage before a salesperson becomes involved.
4. Natural Conversations Instead of Fixed Scripts
One of our main objectives was to avoid creating another rigid, scripted telephone system.
By using OpenAI, the agent can interpret different ways of asking the same question and respond conversationally.
For example:
“I need someone in technical.”
“My phones aren’t working.”
“Can you put me through to IT?”
These requests can be interpreted according to their meaning rather than requiring specific keywords.
The AI can also ask follow-up questions when the caller’s request isn’t clear.
How Is It Hosted?
Our solution uses:
- 3CX – Business telephone system and call handling.
- OpenAI – Conversational intelligence and natural-language understanding.
- Amazon Web Services (AWS) – Cloud hosting for the AI application and supporting services.
- Custom configuration and integration – Developed and configured by Telecoms4Business.
Using AWS gives us flexibility over the hosting environment and allows us to develop additional functionality as the solution evolves.
Why Did We Build It?
We believe AI has the potential to significantly improve how businesses handle incoming telephone calls.
Traditional IVR systems are useful, but they often create unnecessary friction.
Customers don’t necessarily know which department they need. They simply want to explain their problem and speak to someone who can help.
Our aim was to create an AI receptionist that understands the customer’s requirements and takes the most appropriate action.
For businesses handling a mixture of customer service, technical support and sales calls, this could be particularly useful.
What’s Next?
We’re continuing to explore how conversational AI can improve the experience for businesses using 3CX.
Some of the areas we’re particularly interested in developing further include:
- CRM integrations and automated customer record updates.
- AI-generated call summaries.
- More advanced technical troubleshooting.
- Intelligent appointment scheduling.
- Automated support ticket creation.
- Additional integrations with business applications.
Interested in How We’ve Built It?
We’re keen to hear from other 3CX users and developers experimenting with AI-powered call handling.
Has anyone else developed a similar integration using OpenAI and 3CX?
It would be interesting to compare approaches, particularly around call transfers, conversational accuracy, latency and managing more complex customer enquiries.
For anyone interested in our work, the solution has been developed and configured by Telecoms4Business, a UK business telecommunications provider specialising in 3CX, VoIP, connectivity and managed IT.
Website: https://www.telecoms4business.co.uk/
Happy to discuss our approach and exchange ideas with other members of the 3CX community.
Thanks!
Telecoms4Business
