AI · SOFTWARE · CLOUD

AI products and custom software.

Vectorfy is a Melbourne engineering consultancy. We design and build AI products and custom software, connect them to the systems you already run, and operate the cloud infrastructure behind them.

Illustrated Conversation Insights interface showing a call summary, quality highlights and the transcript evidence behind them. Example data.

Featured project · Conversation Insights

See delivered work

Delivered work.

Client deliveries, an internal product, and a tool in development. Each shows the product and the engineering behind it.

ENTERPRISE DELIVERY · CLIENT ANONYMISED

Conversation Insights

Call review for quality, compliance and customer safety. Reviewers read a summary, check the transcript and audio behind each finding, and record their decision.

Fine-tuned transcription and speaker identification
Review interface inside Dynamics 365
GPU scheduling on Kubernetes
GitOps deployment

How it was built Product walkthrough

INTERNAL PRODUCT

Business FinOps

Time, invoicing, bank reconciliation and superannuation records in one application. An assistant prepares drafts that a person reviews before anything is sent.

Xero, MYOB and QuickBooks
Payroll and superannuation records
Bank feeds, documents and email

See the workflow

TOOLING IN DEVELOPMENT

Autoflow

Records how staff actually work across browser portals: the clicks, the fields, the copy and paste, and the API calls behind them. Autoflow turns that evidence into an automation plan an engineer reviews before anything runs.

Authorised, selected-tab capture
Field mapping and data lineage
Reviewed plan: API, browser or hybrid

How capture works

ENGINEERING CASE STUDY

What made the enterprise project difficult.

The constraints were fixed. The system had to work within them.

01
The cloud provider could not change
The mandated provider ran the AI model slowly. We changed the framework, model, prompts and tests, then delivered within the approved environment.
02
Transcription needed a custom model
A leading speech model did not meet the requirements. We fine-tuned a transcription model and built the audio pipeline, including speaker identification.
03
The CRM had to stay
The customer system could not be replaced and its integration was poorly documented. We built the review interface inside it.
04
GPU cost needed control
Custom speech models need GPU computing. We scheduled that work on Kubernetes to reduce running cost, with GitOps managing deployment.

Read the full account

Client name, provider and proprietary methods withheld.

HUMAN REVIEW

Every AI finding links to its source.

Reviewers see the transcript and audio behind each summary, quality check or flagged issue, then record a decision that stays with the case. Access is controlled and every change is logged.

Illustrated product view. Conversation evidence with a source excerpt and reviewer note. Example data.

Illustrated product view. Example data.

Product walkthrough

ENGINEERING CAPABILITIES

AI architecture. Software delivery. Cloud operations.

AI architecture & delivery

Choose and test models for the job, build the application around them, and add human review where decisions need checking.

Software & integrations

Web and mobile applications, APIs, and integrations across accounting, CRM and field-service platforms: Xero, MYOB, QuickBooks, Dynamics 365, Salesforce, HubSpot, Zoho CRM, Zendesk, ServiceNow and Simpro.