Peak call volume
Limited staff during peak hours means missed calls can become lost order opportunities.
CPIN Technology · AI Voice & Enterprise Software
Building Practical AI Software for Real-World Operations.
CPIN Technology develops AI-enabled software, voice workflow automation, and digital products, with AnyTaxi as its flagship mobility technology initiative.
On the ground
Phone orders lose information between steps.
Fleet phone orders span calls, address capture, order entry, and dispatch. When any step requires manual re-entry, errors and delays accumulate.
These are common fleet operating scenarios, not measured results from any specific customer.
Repeated verbal checks
Addresses, times, and notes are often repeated and re-keyed, creating room for disputes.
Fragmented order data
Phone notes, messaging apps, and dispatch systems each hold fragments of the same order.
Hard-to-use operations data
Without structured records, it is difficult to analyse call volume, review rates, and dispatch efficiency.
Flagship product
AnyTaxi turns phone orders into a traceable data workflow.
AnyTaxi starts with the phone call, converts it into a structured order, and connects it to dispatch and fleet operations. The goal is not transcription alone, but a complete path from conversation to operational data.
Order workflow design
5 inspectable stages-
01
Inbound call
A voice workflow guides the caller to provide pickup, time, and trip details.
-
02
Speech-to-text
Call audio becomes text, creating a traceable processing record.
-
03
Field extraction
Address, time, passenger count, and notes are extracted into structured fields.
-
04
Human review
Low-confidence or ambiguous details are confirmed by a person before proceeding.
-
05
Order & dispatch
The order is created and connected to dispatch, producing traceable operations data.
The workflow is used in operational settings. Automation and human intervention vary by deployment.
- Inbound voice orders with call records
- Address, time, and passenger extraction
- Order creation connected to dispatch
- Structured fleet operations data
R&D
Advancing voice AI through verifiable engineering.
We treat voice AI as an operating system, not a single model. Workflow definition, data pipelines, human review, and testing must all be inspectable and adjustable.
See the architecture and milestonesVoice entry
Inbound callVoice workflow
Voice processing
Speech-to-textIntent & field extraction
Operations core
Order serviceDispatch interfaceHuman review
Data & observability
Operations dataEvent logTest samples
Cross-cutting: access control · logging · monitoring · data minimisation
Problem definition & workflow design
Define the order workflow, exception cases, and required fields before choosing technology.
Voice and data pipeline
Build a traceable chain from call audio and speech-to-text to structured field extraction.
Testing and human review
Design review paths for ambiguous details and use test sets to check pipeline behaviour.
Enterprise software
Product-led, with selected enterprise software engagements.
AnyTaxi is our core product. The services below draw on the engineering work behind it and suit projects involving AI automation, platform development, or system integration.
AI & intelligent automation
- Voice AI and conversational workflows
- Document and data processing
- Workflow automation
- Internal enterprise tools
Software & platform development
- Corporate sites and web apps
- Admin back offices
- APIs and databases
- Cloud integration
Digital products & H5
- Mobile interface design and build
- Interactive web experiences
- H5 campaign and product pages
- Product prototypes
Integration & automation engineering
- Cross-system integration
- Deployment automation
- Test automation
- Business operations tooling
Technical notes
Writing down the methods and judgement behind the R&D.
Voice processing
Address Extraction for Voice Orders: From Speech to Dispatchable Fields
Addresses drive dispatch decisions. This note covers common spoken-address variants and a pipeline from speech to human review.
Read the noteSystem design
From Voice to Dispatch: Designing an Event-Driven Order Pipeline
Phone orders are not a single action. This note models voice, review, orders, and dispatch as events, with failure handling and observability.
Read the noteTesting & operations
Observability and Testing for Voice AI Systems
Voice AI takes open-ended audio and produces structured output. Testing needs more than one accuracy number. This note covers test layers, metrics, and privacy.
Read the noteAbout CPIN
Software engineering starts with a concrete problem.
CPIN Technology focuses on AI systems, voice workflow automation, and digital product development. AnyTaxi is already deployed in real fleet operations and continues to be validated and improved rather than remaining a demo.
- PracticalConfirm the problem and workflow before choosing technology.
- VerifiableState capability status and test results separately.
- Data minimisationProcess only what operations require, with access control and records.
Work with us
Planning a fleet pilot, technical partnership, or custom build?
Tell us briefly about the need, context, and timeline. The relevant contact will reply.