Datics AI
← All work

Sentrics

A senior living app's team gets proven voice and device messaging on AWS

Sentrics, a senior living technology company, had a live Android app for its Engage360 product and two AWS integrations its team could not finish. We proved voice to Amazon Lex and Transcribe and device messaging over MQTT, then documented every step.

Healthcare & WellnessAndroid and AWS integration modulesAI added to a live product

The challenge

The client's team was stuck on two AWS integrations for its Android app: recorded audio to Amazon Lex and Transcribe, and messages to a device over MQTT.

What we did

We built a small reference app for each integration, proved it end to end, then wrote down the AWS setup and recorded a configuration and testing video.

The outcome

The team received both integrations working, with setup it could repeat in its own AWS account. The CTO rated both jobs five stars.

What shipped
Two reference modules with AWS setup guides
Voice
Android audio to Amazon Lex and Transcribe
Messaging
Android and Node.js clients over AWS IoT MQTT
Rating
Two 5 star reviews from the client's CTO

The starting point

Sentrics builds technology for senior living, including the Engage360 resident engagement product with an Android app. Its own engineering team owns and runs the product.

What had to change

The team could not get recorded audio from its Android app into Amazon Lex and Amazon Transcribe, or push messages over AWS's MQTT service to an Android app on a dedicated device. The CTO called the first task very difficult.

What could not break

  • The client's team keeps ownership of its Android app and codebase
  • Everything runs in the client's own AWS account
  • Each job has a clear finish line: a working module plus its setup

The calls we made

We treated each integration as something to prove end to end on its own, then hand over so the client's team could own it.

  1. 01

    One small reference app per integration

    A self contained app proves the AWS setup works without the noise of the full product, so the team can see exactly what to copy into its own code.

    Ruled out

    Changing the client's full app ourselves, which would have left the team without a clear pattern to own.

  2. 02

    Let AWS do the speech work

    Speech recognition and intent handling run on Amazon Transcribe and Amazon Lex, so we put our effort into audio capture and transport, the part the team was stuck on.

    Ruled out

    Building custom speech or intent models.

  3. 03

    Write down and record the AWS setup

    AWS configuration is where the team had been blocked, so we documented the whole flow for both apps and recorded a configuration and testing video.

How the system fits together

audiorecorded audioaudio forintentpublishmessagereplypattern to adoptpattern to adoptResident or staffmemberSpeaks into the AndroidappVoice moduleAndroid app that recordsand sends audioNode.js clientPublishes and receivesMQTT messagesAmazon TranscribeSpeech to textAmazon LexIntent from the requestAWS IoT CoreMQTT message brokerDevice messagingmoduleAndroid app on a dedicateddeviceEngage360 Android appOwned by the client's team
PeopleBuilt by DaticsAI modelOutside serviceClient's existing systemWe built the capture and transport layer; AWS ran speech and messaging in the client's own account, and the modules showed the team what to fold into its app.

What we built

Voice module

An Android app that records audio and sends it to Amazon Lex and Amazon Transcribe, proving the voice path end to end.

02

MQTT messaging module

A simple Android app and a Node.js client that send and receive messages over AWS's MQTT service, for an Android app running on a dedicated device.

03

Handover kit

Written documentation of the full AWS flow for both apps and a detailed configuration and testing video the team could follow.

How it works

Capture

The Android app records audio from the user.

Recognize

Audio goes to Amazon Transcribe and Amazon Lex, which return text and intent.

Message

A Node.js client publishes a message over AWS MQTT.

Receive

The Android app on the device receives the message and can reply over the same channel.

What changed

The client's team received both integrations working end to end, with setup it could repeat in its own AWS account and app. The CTO rated both jobs five stars and came back for the second job after the first.

Under the hood

Datics role
AI added to a live product
AWS services
Amazon Lex, Amazon Transcribe and AWS IoT Core MQTT, configured in the client's own AWS account.
Boundary
AWS runs speech recognition and intent handling; we built the capture and transport layer, not the language models.
Handover
Reference apps, written AWS flow documentation and a configuration and testing video for the client's team.

Cloud and infrastructure

Amazon LexAmazon TranscribeAWS IoT Core

Apps and APIs

Node.js

Also in the stack

AndroidMQTT

How it shipped

  1. Voice first

    A short fixed scope job: the Android voice module, working against Amazon Lex and Transcribe.

  2. Device messaging

    A second job: the Node.js client and Android app exchanging messages over AWS IoT MQTT.

  3. Handover

    The AWS flow for both apps documented step by step, plus a recorded configuration and testing walkthrough.

Short jobs with a clear finish line

Each integration was a fixed scope job that ended with a working module and its documentation, not an open ended engagement.

Reviewed before handover

Our CTO reviewed the work before it reached the client's team and stayed the client's single point of contact.

What this unlocked

The client's team now has a proven, documented path for adding voice and device messaging to the app it already runs.

Related work

Have a product with the same kind of constraints?

Tell us what already exists and what you want it to do. We will tell you what it takes.

Book a 30-minute Product Review