AI Engineering · India
Accurex - AI Workflows for Field Sales Attendance and Call Evaluation
Two automated workflows. One checks field sales attendance. The other turns tele-caller audio files into transcripts, with speakers identified and non-English conversations translated into English.
Discuss Your AI Project- Attendance Automation
- Call Transcription
- Speaker Identification
- English Translation
About the client
Accurex is an operations-driven company that manages distributed teams and a high volume of conversations between its employees and customers. It wanted to cut out manual processes, improve accuracy, and get structured information from its day-to-day operational data.
- Industry
- Medical Devices
- App type
- AI-Powered Attendance Automation and Call Evaluation System
- Practice
- AI Engineering (Build)
- Services
- Agents & Agentic Workflows, AI Platform Partnerships & Professional Services, AI Strategy, Governance & Compliance Advisory
Tech stack
- Gemini models (transcription, speaker identification, translation), AI-based OCR and image processing, Excel and spreadsheet pipelines, and Shared drive integration.
Engagement summary
We built two automated workflows for Accurex. One handles attendance for field sales staff. The other processes audio files of calls made by its tele-callers.
For attendance, the workflow reads each spreadsheet entry and checks that the required fields are filled in. It reads the date and time from the IN and OUT images, works out how long each person worked, and marks them Present or Absent against the working-hour requirement. Records with problems go to an error log with the reason for rejection.
For calls, the workflow picks up AAC audio files from a shared drive. It transcribes each call, identifies who is speaking, translates the conversation into English, and saves the result in an Excel sheet ready for review and analysis.
Challenges faced by the client
Accurex relied on manual and semi-automated workflows, which caused errors and delays.
Manual attendance tracking
Field sales staff submitted geo-tagged, timestamped selfies. Attendance was then handled through image uploads and spreadsheets, which led to errors and delays.
No automatic validation
Missing or incorrect attendance records weren’t caught automatically.
Attendance couldn’t scale
Processing attendance for more teams meant more admin overhead.
Calls reviewed by listening
Accurex’s tele-callers phone diagnostic labs through a phone-based CRM. Their managers had to listen to each call to rate them. There was no standard transcription, no speaker identification, and no automated translation for calls in other languages.
Heavy reliance on people
All of this depended on manual review, which pushed up cost and turnaround time.
Together, these issues affected efficiency, accuracy and how quickly decisions could be made.
Results
Attendance processing is 100% automated, with zero manual validation
Attendance is generated every day without admin intervention
Duplicate and inconsistent records are eliminated
Every call goes through the same evaluation process
Structured Excel outputs help the team make decisions faster
The system can handle more volume without adding headcount
How it works
- Records missing required details are separated out and logged, so only valid records reach the hours calculation.
- Processed records are cleared after each run so nothing is counted twice, and the workflow runs daily without anyone starting it.
- Each call comes out as a transcript with speakers identified and non-English conversations translated, in one Excel sheet.
Core solution
Read the data
Attendance entries are pulled from the Excel spreadsheet.
Check the required fields
Valid records move on to the calculation. Invalid ones go to an error log with the rejection reason.
Read the IN and OUT images
AI reads the date and time from each image.
Calculate and compare
The working duration is worked out and compared with the predefined benchmark.
Mark attendance and clear records
Each person is marked Present or Absent, and processed records are cleared to prevent duplicates. The workflow runs daily without manual triggers.
Pick up the audio
AAC audio files are taken directly from the shared drive.
Transcribe and identify speakers
Gemini models turn the speech into text and tell the speakers apart.
Translate
Gemini’s translation converts the conversation into English.
Structure and export
The output is formatted with Gemini-assisted processing and exported to Excel for analysis and reporting.
More Case Studies
AI Engineering · IndiaPolicyBoss - A Voice AI Agent for Motor Insurance Policy Verification
Insurance / InsurTechA voice AI agent that supports policy verification calls alongside live agents, retries unfinished verifications, and follows up on WhatsApp after every attempt.Read the Full Case Study
Software Engineering · IndiaJio Media Stack
Jio · Digital CommunicationWe developed a bespoke Video Calling SDK for a leading communication technology client, eliminating third-party dependencies and enhancing control over their infrastructure in the digital communication industry.Read the Full Case Study
AI Platforms & Infrastructure · United StatesSecure Cloud DMS Stack
PropTechWe transformed a legacy on‑prem shared drive into a secure, cloud‑based document management system with granular roles, rich collaboration, and seamless access across locations for a fast‑growing PropTech leader.Read the Full Case Study
Let’s Explore How We Can Build Together
Whether you’re building a new product or modernizing an existing one, get insights from our product, engineering, and AI experts.