PROJECT

Conversational AI for Voice, Chat, Sales, and Support

Custom software development, AI voice agents, Chat automation, Speech-to-text, Text-to-speech, CRM integration, Calendar integration, Call analytics

  • Industry: Business services
  • Norway Location: Norway
  • Delivery time: 16 weeks
  • Technologies: JavaScript / TypeScript (Node.js), Python
4+ months

This client is a call center from Norway, where staff handle technical and general service conversations for customers. Calls that require routine answers, lead qualification, appointments, or escalation take staff time. The conversational AI software provides voice and chat agents for inbound and outbound conversations, answers questions from approved business knowledge, and records outcomes for support and sales teams.

Summary

Problem

The call center spends staff time on routine support, sales qualification, appointment requests, and call reviews.

Solution

The solution became a conversational AI workspace with voice and chat agents. It connects business knowledge, transfers, and call records.

Result

The software is made to reduce manual work across support and sales conversations while keeping important outcomes available for review.

Client story

The client is a call center from Norway. It handles inbound technical and general service calls, as well as sales and lead-generation conversations.

Routine questions, qualification calls, and appointment requests take staff time. Escalations also require staff involvement.

The request was simple: provide real-time voice agents for inbound and outbound calls, answer routine questions from approved knowledge, connect conversations with business systems, and preserve call records.

The solution was shaped as separate workspaces for live support, lead qualification, and appointments. Knowledge management and call history were provided as dedicated workspaces.

Voice connections and business rules were kept in one modular system.

The client asked for

  1. Provide real-time voice agents for inbound and outbound calls.
  2. Answer routine questions from approved business knowledge.
  3. Connect conversations and actions to business systems.
  4. Store recordings, transcripts, outcomes, and usage metrics.

Why ready-made software did not fit

Ready-made products cover parts of the task. Voice platforms provide conversation runtime, while call-recording tools focus on recording and transcription. The cited products do not describe the full combination of business knowledge, lead qualification, appointment operations, live transfers, CRM connections, and administration required by this call center.

Project overview

The solution was designed around the call center’s main conversation paths: support, sales, and appointments. Conversation records and business connections are shared across the system.

Real-time voice agents handle inbound and outbound calls through telephony, SIP, VoIP, or browser voice connections. Speech-to-text turns spoken language into text, while text-to-speech produces natural audio responses.

Knowledge was separated by store or client. FAQs, product information, and property information are managed in the administration screen.

TXT skill files, greetings, and business hours are kept with the agent settings.

To handle business actions, the system was connected with CRMs, ticketing systems, and calendars. Email, messaging, webhooks, and REST APIs support additional connections.

Agents can qualify leads, manage appointments, and transfer calls to live representatives.

Special attention was paid to the records created during conversations. Recordings, transcripts, and summaries are available in a searchable history, together with outcomes and usage metrics.

Key features delivered

  • Real-time inbound and outbound voice agents
  • Speech-to-text and text-to-speech conversations
  • FAQ and knowledge-base answers
  • Live intent detection and call transfers
  • Lead qualification with scores and outcomes
  • Appointment scheduling, rescheduling, and cancellation
  • CRM, calendar, email, messaging, webhook, and REST connections
  • Call recordings, transcripts, summaries, and logs
  • Separate knowledge for stores or clients

Can be added

+ Outbound qualification campaigns + Browser voice connections + Usage controls + Follow-up automation

Who else can use this solution

Other businesses with a similar conversation task can adapt the same solution to their own knowledge, channels, rules, and business systems.

A sales and lead-generation operation

Task It runs inbound or outbound conversations to qualify leads.
How the solution helps The solution scores leads, records outcomes, and connects follow-up actions with business systems.
Business services

An appointment-based service business

Task It handles appointment requests and calendar-based follow-ups.
How the solution helps The solution schedules, reschedules, or cancels appointments and keeps the confirmation with the conversation record.
Home services

A multi-store organization

Task It needs separate agents with isolated knowledge for each store.
How the solution helps The solution keeps store-specific information, agent settings, phone numbers, and records separated.
Retail stores

Technology stack

JavaScript / TypeScript (Node.js)

Application services

Node.js services connect the conversation experience with channels and business integrations.

Python

AI and data services

Python services support speech, calendar, document, reporting, and data-processing functions.

@telnyx/webrtc

Browser voice calls

This WebRTC client connects browser voice and SIP-based communication.

jssip

SIP communication

This JavaScript SIP library supports browser-based SIP and WebRTC call connections.

rrule

Recurring appointments

This library handles recurring calendar rules for appointment workflows.

icalendar

Calendar events

This Python library reads and creates iCalendar events for scheduling.

csv-parse

CSV data reading

This parser reads CSV data used in imports and reports.

PDF libraries

PDF document handling

Libraries such as pymupdf and pypdf extract, split, merge, and process PDF documents where needed.

Integrations

Telephony, SIP, and VoIP

The solution exchanges live call audio through telephony, SIP, VoIP, or browser voice connections.

CRM systems

Conversation data, lead outcomes, and follow-up actions move between the agents and CRM systems.

Google Calendar

Appointment requests and calendar availability are exchanged with connected calendar systems.

Webhooks and REST APIs

Conversation events and business actions are sent to external systems through webhooks or REST APIs.

Email and messaging

Incoming messages and follow-up communication can connect to email and messaging systems.

The result

Support agents see the live conversation clearly

The live conversation workspace brings caller speech, detected intent, caller identification, and transfer status together. The call center can keep routine conversations with the agent and pass a call to a live representative when required.

The transcript gives the support team a readable record while the conversation is in progress.

How it works

  1. Speech-to-text converts caller speech into text in real time.
  2. The system detects the conversation intent and shows the caller phone number.
  3. The workspace shows the transfer decision and supports transfer to a live agent.
Tip: The same conversation can remain available for later review through the call history.
Live conversation workspace

Live conversation workspace

Functions on this screen
Implement speech-to-text functionality Real-time Speech Recognition Caller ID Integration Transfer calls to live agents Detect intents

Sales teams focus on qualified leads

The qualification queue turns inbound and outbound conversations into a clear list of lead outcomes. Sales representatives see the lead score, summary, and follow-up record together.

This keeps qualification information close to the next sales action instead of leaving it only in a call recording.

How it works

  1. The system lists lead details and qualification status.
  2. The workspace shows the lead score and conversation summary.
  3. The call and follow-up record stores the qualification outcome.
Lead qualification queue

Lead qualification queue

Functions on this screen
Lead Qualification Automation Call Logging and Follow-Up

Appointment requests stay connected to the calendar

The appointment panel connects a conversation with calendar availability. It keeps caller details, the requested appointment, confirmation status, and the relevant scheduling action in one place.

The workflow covers scheduling, rescheduling, and cancellation for service conversations.

How it works

  1. The panel shows caller details and the requested appointment.
  2. The system displays available calendar time slots.
  3. The appointment status and confirmation update for schedule, reschedule, or cancellation actions.
Appointment scheduling panel

Appointment scheduling panel

Functions on this screen
Appointment Scheduling Tool

Each agent uses the right business knowledge

The configuration workspace keeps FAQs, product or property information, store details, and skill files under the correct store or client. Administrators also manage greetings, business hours, and agent settings.

Separating this information helps the call center maintain distinct knowledge for different clients or stores.

How it works

  1. The editor stores FAQs and product, property, or store information.
  2. The system lists TXT skill files for the selected store or client.
  3. The agent configuration keeps greetings, business hours, and agent settings together.
Note: Knowledge can be maintained separately for each store or client.
Knowledge and agent configuration

Knowledge and agent configuration

Managers review calls and outcomes in one history

The call history makes completed interactions searchable by caller ID, date, duration, and outcome. Managers can open the recording and transcript alongside analytics and usage metrics.

This gives the operations team a consistent place to review support and sales conversations.

How it works

  1. The system lists completed calls with date, duration, and caller ID.
  2. The recording player provides the stored audio and the transcript viewer shows the text.
  3. The outcome and analytics panel displays call results and usage metrics.
Call history and outcome report

Call history and outcome report

Functions on this screen
Conversation logging Call Logging and Follow-Up

What the solution gives

With this solution, the call center can:

  • Handle inbound and outbound conversations with voice agents
  • Answer routine questions from controlled business knowledge
  • Transfer calls to live representatives when needed
  • Qualify leads and keep scores, outcomes, and follow-ups together
  • Schedule, reschedule, and cancel appointments through calendar connections
  • Review recordings, transcripts, summaries, and usage metrics

Could ready-made systems do the job?

Before the design, the task was compared with ready-made products by their own descriptions.

What the client needs Voice agent platforms Amazon Lex, Callab AI, VoiceDock Call recording tools GreenWeka Custom solution
Real-time inbound and outbound voice agents —
Answer from business knowledge partly —
Connect actions to business systems partly —
Store recordings and conversation records partly
Fits the client's own rules: stages, roles, reports partly —
All steps in one place — —
No manual steps between programs partly —

By the public descriptions of the products and their category.

Ready-made products cover real parts of the task: voice platforms handle live conversations, and call-recording tools preserve and summarize calls.

A ready-made product is enough when the call center needs only one of those parts. This client needed business knowledge, qualification, appointments, transfers, integrations, administration, and review in one connected system.

The parts and the client’s own rules would otherwise remain spread across products, with manual steps between them.

In the end

In the end, the solution was made to take routine manual work out of the path from a customer conversation to a clear business outcome.

FAQ

Thinking about building a similar system? Read the answers below to see how conversational AI can connect voice, chat, business knowledge, and operational actions.

Need something similar? Let's talk
Or browse more projects