Case Studies | Retail

Answering every call without tying up staff

Voice AI Concierge and Ordering System

About

A Retail business working with Cloud Combinator on AWS. The client is anonymised at their request.

Challenge

The challenge had four focus areas.

Phones tie up staff

Inbound calls for new orders, order tracking and general questions constantly interrupt the team during service. Automating these calls, while still sounding like a helpful member of staff, was the core aim: reduce workload without reducing customer satisfaction.

Voice has to be fast and natural

On a phone call, delay and stilted speech break the experience immediately. The agent needed to respond with less than three seconds of latency per turn and speak with natural, empathetic text-to-speech, capturing intent the moment the caller pauses to minimise dead air.

Orders must be accurate and safe

Real orders mean real money, so the agent had to process menu modifiers, upsells and dynamic pricing from the platform backend, verify addresses and delivery zones, and generate payment links, all behind strict guardrails that prevent price manipulation and block order lookups via spoken phone numbers.

Knowing when to hand off

No automated system should trap a frustrated caller. The agent needed an escalation engine that detects frustration or an explicit transfer request and hands the call to venue staff cleanly, with a whisper announcement so the human picks up in context.

Solution

A call arrives through Amazon Connect, which acts as the cloud contact centre and telephony carrier. Connect triggers a Lambda that looks the caller up against the platform backend and, in parallel, fetches the venue's voice menu, its information and any active orders, caching them in Amazon DynamoDB so the rest of the call stays low-latency. A dynamic greeting then routes the caller into the appropriate conversation flow.

< 3s

Target latency per conversation turn for a natural phone experience

4

Core call flows: order tracking, enquiries, new orders, human handoff

22-day

Engineering build, deployed as CloudFormation infrastructure-as-code

By the numbers:

  • < 3s - Target latency per conversation turn for a natural phone experience
  • 4 - Core call flows: order tracking, enquiries, new orders, human handoff
  • 22-day - Engineering build, deployed as CloudFormation infrastructure-as-code
Changes

The engagement delivers a working voice AI concierge that identifies callers, retrieves active orders, presents the menu and takes new orders across four defined conversation flows, with escalation to humans when needed and guardrails protecting pricing and caller data. Acceptance is on demonstrating those flows working end to end within the target latency, with the security guardrails in place.

  • Natural, low-latency voice — Amazon Connect, Lex V2, Polly neural TTS and Bedrock Nova Lite combine for a responsive, natural conversation under a three-second-per-turn target.
  • Accurate ordering — The agent handles menu modifiers, upsells and dynamic pricing from the platform backend, with address and delivery-zone checks via the Google Places API.
  • Safety guardrails — Caller-ID authentication, deterministic pricing from cached menu data and a ban on AI-generated discounts protect both the venue and the customer.
  • Human handoff — An escalation engine detects frustrated callers or explicit transfer requests and routes to venue staff with a whisper announcement.
  • Handover — AWS CloudFormation templates, documentation and Loom walkthroughs leave the client able to deploy, operate and extend the system independently.

With a voice AI concierge now able to answer, understand and act on inbound calls, the client can offer its food venues a way to capture every order without adding to the workload during service. Built serverless and deployed as infrastructure-as-code, the system is ready to roll out across venues and to grow into whatever the platform takes on next.

AWS Stack

Amazon Connect

For cloud telephony, DID phone numbers and call routing.

Amazon Lex V2

With streaming transcription for real-time speech recognition.

Amazon Polly

Neural text-to-speech for natural, empathetic spoken responses.

Amazon Bedrock

With Nova Lite for conversational understanding and order building.

AWS Lambda

For integration with the platform backend, Google Places and session logic.

Amazon DynamoDB

For low-latency caching of tenant, menu and conversation state.

AWS End User Messaging

For order confirmation and payment-link SMS.

AWS CloudFormation

For automated deployment of the whole stack.

YOU MIGHT LIKE

Related success stories

View all case studies

Case Studies | Insights

Utilising Language Recognition, Speed, and Enhanced Security to Make Social Media a Force for Good

  • Here, we take a detailed look at how the Cloud Combinator team collaborated with another cutting-edge AI service provider that provides intelligent systems to “make social media more social” for brands and users alike.
  • Arwen AI is a UK-based startup specialising in AI solutions to manage and enhance brands’ social media interactions. Founded in 2020 by Matt McGrory, Dr. David Cole, and Joel Bailey, Arwen. AI focuses on using AI to automatically detect and remove spam, toxic comments, and other unwanted content from social media platforms.
  • The team at Arwen have three core products. ‘Moderate’ is focused on identifying and removing toxic content from social media channels. ‘Engage’ helps brands identify and engage with meaningful conversations on social media, and ‘Customize’ allows brands to apply bespoke algorithms to their channels - creating an even more effective moderation and engagement.
Read more
CONTACT US

Ready to turn AI into impact?

We'll help you spot the highest-value opportunities, reduce risk around your first AI initiative, and define a clear path to results from day one.

Why talk to us:

Outcome-driven recommendations

AWS-recognised delivery expertise

Risk-aware AI adoption

Clear next step, not a sales pitch

Start with a focused 20-minute conversation about your goals — no pressure, no commitment.

We use cookies

This website uses cookies to enhance user experience and to analyze performance and traffic on our website.

See our Privacy Policy for details.