Cognify Labs Pty Ltd · Sydney, Australia

Building practical AI products for real-world problems.

Cognify Labs is an Australian AI product company developing intelligent software that transforms complex workflows into simpler, faster and more accessible experiences.

4 products, all deployed · public demos for two · built and run on AWS in Sydney

Product portfolio

Four products, each aimed at a specific problem.

We design, build and operate our own software. Each product below has its own domain and a stated stage. Screenshots come from the products themselves, and every caption says what data is shown.

AdviceGenie practice dashboard showing open advice cases, product-intelligence counts and the pipeline from product documents to an evidence-backed Statement of Advice
Practice dashboard in the public AdviceGenie demo. All clients, advisers and products shown are fictional.

AdviceGenie

Prototype · in development
adviser.advicegenie.com.au · advicegenie.com.au

An advice platform for financial professionals. It turns product documents into structured, cited data that supports research, modelling and Statements of Advice.

For
Australian licensed financial advisers, paraplanners and practices.
Problem
Researching products, checking them against an approved product list and documenting advice is slow, manual and hard to evidence.
  • Product intelligence: PDS, TMD and fee documents become structured, version-controlled product records.
  • Evidence-backed recommendations: each one links to the documents behind it.
  • Adviser in control: every recommendation must be approved, modified or rejected, and the reason goes into an audit trail.
  • Retirement projection engine, compliance and consistency checks, SOA assembly and a client portal.
Where AI fitsThe demo shows the planned workflow from document ingestion to adviser approval. It runs entirely in the browser on fictional data. Its AI steps are simulated and its figures are computed, not hard-coded. The production platform is not yet generally available. It helps licensed advisers and does not give advice to consumers.

Also live: advicegenie.com.au. It's a free consumer site with plain-English general information and calculators covering super, insurance, investment, SMSF and estate planning.

Assetwise dashboard showing 16 assets, their purchase and current value, and an insurance cover position comparing recorded value with sums insured
Assetwise dashboard from the public demo. It shows a fictional household that lives only in your browser.

Assetwise

Live · early access
assetwise.advicegenie.com.au

An intelligent asset register. Receipts, invoices, warranties and insurance policies become one organised, searchable record of what you own.

For
Australian households and property investors, and later the advisers and brokers who serve them.
Problem
Proof of purchase, warranties and policy details are scattered across inboxes and drawers. They're often missing when you need to make a claim.
  • Document extraction: scanning a receipt or invoice gives a structured asset record. Each field carries a confidence score and the page it came from.
  • Expiry tracking: warranties, return windows, renewals and service dates are worked out from the documents, with reminders.
  • Insurance information: certificates and PDSs are read into clauses with page citations. Recorded values are compared with sums insured.
  • Claims records, property cost records, household members, CSV exports and optional two-step sign-in.
Where AI fitsAmazon Textract reads the document. A Claude model on Amazon Bedrock extracts and classifies it. Deterministic rules then check the result, and the user confirms before anything is filed. Personal details are masked before any text reaches the model. An assistant for questions about your records is built but not yet released, so it's labelled "coming soon".
RateScan homepage showing today's median owner-occupier variable, investment variable and personal loan rates across 83 lenders
RateScan's public homepage, captured 28 October 2026. The figures are that day's published benchmarks.

RateScan

Live
ratescan.com.au

Presents Australian lending rates clearly, so borrowers and brokers can see where a loan sits against the market.

For
Borrowers checking their rate, mortgage brokers, and lenders benchmarking their position.
Problem
Advertised rates are spread across over a hundred institutions, and it's hard to tell whether your own loan is still competitive.
  • Data source: public product data published by lenders under the Consumer Data Right (Open Banking).
  • Coverage & frequency: benchmarks drew on 83 lenders with qualifying products on 28 October 2026. Rates refresh once each weekday morning, not in real time.
  • Median and quartile benchmarks across nine categories, plus a daily rate-change monitor.
  • Rate Tracker with email alerts, repayment and switching calculators, and a lender directory.
Where AI fitsThe rate figures are computed by a conventional data pipeline. A Claude Haiku model on Amazon Bedrock then writes a plain-English daily market briefing from that day's computed figures and changes.
ATS dashboard with open roles, candidate totals, a candidate funnel chart and a time-to-hire trend
ATS recruiter dashboard, shown with the platform's built-in sample data. No real candidates appear.

ATS Platform

In production use
Deployed for Advice Lab · ats.advicelab.com.au

An applicant tracking and recruitment platform that replaces spreadsheet hiring with one managed pipeline, from job posting to hire.

For
HR teams and hiring managers at small and mid-sized organisations.
Problem
Recruitment run from spreadsheets and inboxes loses candidates, history and accountability.
  • Candidate management: public careers page, multi-step applications with CV upload, candidate profiles, tags and scores.
  • Workflow: drag-and-drop pipeline from Applied to Hired, interview scheduling on a calendar, and panel feedback.
  • Automation: automatic candidate emails when an application is received, an interview is scheduled or the stage changes.
  • Reports, a full audit trail, role-based access and admin user management.
Where AI fitsNone yet. The current platform is conventional workflow software. Automated CV parsing and AI-assisted screening are on the roadmap, and we don't claim them today.

The platform is private to the organisation using it, so there's no public demo. Our first version was built and delivered in about two weeks.

Problems we solve

Complex information, made usable.

Each product starts from a specific problem: important information trapped in long documents, scattered data or manual processes.

ProductThe problemWhat the product does
AdviceGenieAdvice files are built by hand from hundreds of pages of product documents, and every recommendation has to be justified.Turns product documents into structured data and links each recommendation to its evidence. The adviser stays the decision-maker.
AssetwiseHouseholds can't easily prove what they own or what they paid, or see whether it's still under warranty or insured.Builds the asset register from the documents people already have, then tracks the dates and cover that matter.
RateScanBorrowers rarely know when their rate stops being competitive. Comparing products across lenders is tedious.Benchmarks the whole published market every weekday, so a loan can be scored against comparable products.
ATS PlatformSpreadsheet recruitment loses candidates and history, and nobody can see where each hire is up to.One pipeline for every role, with candidate emails sent automatically and a record of every change.
How we apply AI

Foundation models inside conventional, checkable software.

We build AI-powered applications and workflows. We don't train our own foundation models. We use Anthropic Claude models through Amazon Bedrock, combined with OCR, deterministic rules, validation and human review. The flow below is how Assetwise handles a document in production.

01 · INPUT

Document arrives

A receipt, invoice, insurance certificate or PDS, uploaded or photographed.

Private, encrypted S3 · malware scan
02 · READ

Text extraction

Scanned pages become text, with page numbers kept for citations.

Amazon Textract
03 · UNDERSTAND

Model extracts fields

Personal details are masked first. The model returns structured JSON with a confidence score and source page for each field.

Claude on Amazon Bedrock
04 · VALIDATE

Rules check the output

Schema checks, deterministic backstops (dates, totals, sums insured) and duplicate detection catch what the model gets wrong.

Tested rule engine
05 · DECIDE

A person confirms

The user reviews every field against the original before it's filed. Nothing is filed silently.

Human in the loop

Document processing & extraction

Reading receipts, invoices, certificates and long product disclosure statements into structured records with page citations.

ASSETWISE · ADVICEGENIE (DEMO)

Workflow automation & data transformation

Scheduled pipelines that fetch, normalise and publish data, plus event-driven emails and reminders.

RATESCAN · ASSETWISE · ATS

Analysis & decision support

Rate benchmarks, insurance cover comparisons and evidence-backed recommendations. They inform a person's decision; they don't replace it.

RATESCAN · ASSETWISE · ADVICEGENIE (DEMO)

Generated plain-English summaries

Models write readable narrative from figures our code has already computed, such as a daily rate briefing or career research.

RATESCAN · CAREER ADVISER

Models plus conventional software

Every model call sits inside typed contracts, retries, fallbacks and tests, so a bad response fails safely.

ALL AI FEATURES

Secure APIs & integration

Authenticated APIs, role-based access and integrations with public data sources such as the Consumer Data Right and CRICOS.

ALL PRODUCTS
Human oversightA user confirms extracted records before they're filed. In AdviceGenie, an adviser decides every recommendation, and the reason is recorded.
Cited, checkable outputsExtracted values carry their source page and confidence score. Low-confidence results ask for confirmation instead of deciding.
Privacy by designAssetwise masks personal details before any text is sent to a model and restores them in the reply. Documents are stored privately and encrypted.
Data kept in AustraliaBackends run in AWS Sydney (ap-southeast-2). Assetwise and RateScan call models through Bedrock's Australian inference profile.
Evaluation and testsIn Assetwise, extraction and cover-check logic is tested against fixed fixtures, and the server and browser implementations are checked against each other. A rule change that breaks extraction fails the tests.
Honest limitsWe label features that aren't released yet, keep simulated AI out of claims about the live products, and never present outputs as financial, tax or legal advice.
Product demonstrations

See the products working.

Two of the products have full interactive demos you can open without an account. The demos use fictional data and keep it in your browser.

Assetwise review screen: a tax invoice for an espresso machine, with supplier, ABN, model, serial number, price, GST and warranty extracted, each with a confidence score and page reference

Assetwise: a receipt becomes an asset record

  1. Upload a sample tax invoice.
  2. The supplier, ABN, model, serial number, price and warranty are extracted, each with a confidence score and page number.
  3. A stated two-year warranty replaces the 12-month default, and the dates are calculated.
  4. You confirm. The asset, its warranty and its reminders are filed together.

The public demo runs a browser-based extractor on sample documents. Signed-in accounts use the Textract and Bedrock pipeline shown above.

AdviceGenie adviser approval screen listing AI-proposed recommendations, each with evidence links and Approve, Modify and Reject buttons

AdviceGenie: the adviser approves every recommendation

  1. Open a fictional client case and its fact find.
  2. Products are screened against an approved product list, with the exclusions recorded.
  3. Each recommendation links to its evidence and must be approved, modified or rejected.
  4. An approved recommendation flows into the Statement of Advice, and every step lands in the audit trail.

Demonstration prototype: fictional clients, providers and products, with AI steps simulated in the browser.

Development progress

Where each product is today.

This is an early-stage company, and this table shows each product's real stage. "Next" lists work that's already designed or partly built. It isn't a promise of dates.

ProductStageLive todayNext milestones
AdviceGeniePrototypePublic interactive demo of the end-to-end advice workflow (fictional data). Adviser site with early-access sign-up. Consumer education site with calculators.
  • Production backend for product-document ingestion
  • Early-access pilots with advice practices
AssetwiseEarly accessProduction platform with sign-in, the Textract and Bedrock document pipeline, reminders, paid annual plans and a public demo.
  • Release the records assistant (built, not yet switched on)
  • Adviser portal for consented, read-only client access (built, switched off)
  • Email receipt ingestion
RateScanLiveWeekday market benchmarks, rate-change monitor, Rate Tracker with alerts, calculators, AI daily briefings, and a Free plan plus a paid Pro plan.
  • Competitive-analysis dashboard as a paid feature for brokers and lenders
  • Bulk loan-book scoring for enterprise customers
Career AdviserPilotPublic site with career research profiles written with Amazon Bedrock and course data from the national CRICOS register. Open for general feedback.
  • Gather pilot feedback from students and families
  • Seeking founding partners for the next stage
ATS PlatformIn useIn production for one organisation (Advice Lab), covering jobs, applications, pipeline, interviews, notifications, reports and audit.
  • Automated CV parsing
  • Job board integrations
  • AI-assisted screening
What's realEvery product above runs on deployed infrastructure. The live products process real data, and the demos are labelled as demos.
Commercial stageAssetwise and RateScan have paid plans and payment processing live. Both are early in commercial use, and we don't publish customer numbers yet.
What we don't claimNo testimonials, revenue figures, partnerships or certifications appear here, because we'd only show ones we can verify.
Founder and company
David Karunaratne, founder of Cognify Labs

David Karunaratne

FOUNDER · COGNIFY LABS PTY LTD

David founded Cognify Labs and designs, builds and runs its products. He has 25+ years across technology, finance and telco, designing, building and shipping production systems for financial-services firms, startups and enterprises.

That experience includes managing enterprise platforms (performance, security, processes and finance) and more than ten years designing, building, deploying and managing cloud-native applications. Three of the four products serve financial services, where much of that experience was gained.

25+ yrs technology Data & AI Financial services Enterprise platforms Cloud-native · 10+ yrs
Technology and infrastructure

Serverless on AWS, defined as code.

Every product runs serverless on AWS in the Sydney region and is defined as code with AWS SAM and CloudFormation. Each product keeps its own isolated database and API. Shared code, such as authentication, roles and data access, is maintained once and reused.

NeedAWS services in use todayUsed by
Web hostingAmazon S3, CloudFront, Route 53, Certificate ManagerAll products
APIs & computeAPI Gateway (HTTP APIs), AWS Lambda (Python, Arm)All products
Databases & storageAmazon DynamoDB; Amazon S3, with private KMS-encrypted document buckets for AssetwiseAll products
AI inferenceAmazon Bedrock (Anthropic Claude models)Assetwise, RateScan, Career Adviser
Document processingAmazon Textract, GuardDuty malware scanning for S3Assetwise
Background workloadsAWS Step Functions, EventBridge SchedulerRateScan, Assetwise
Analytics dataAmazon Athena, AWS Glue, Apache Iceberg on S3RateScan
AuthenticationAmazon Cognito, with optional authenticator-app MFAAssetwise, ATS, others
Email & monitoringAmazon SES, Amazon CloudWatchAll products

Planned next

  • AdviceGenie production platformMoving from the browser demo to a backend: document ingestion, model inference over long product disclosure statements, and versioned product data.
  • Assetwise assistantQuestions answered from a household's own records, with citations. The Bedrock endpoint is deployed and the user interface is the remaining step.
  • More AI document workCV parsing in the ATS and broader document coverage in Assetwise.
Why cloud and AI compute matter hereModel inference is a direct, per-document cost. A long insurance PDS takes many model calls to read. Getting the AdviceGenie platform and the Assetwise assistant to production will need more Bedrock and Textract capacity, plus separate development and test environments to evaluate model changes safely.

Cognify Labs is an AWS customer. Naming AWS services doesn't imply any partnership with or endorsement by Amazon Web Services.

Commercial direction

Who each product is for, and how it earns.

Where a pricing model is already live, it's stated here. Where it isn't settled yet, we say so.

Assetwise

Early access
Customers
Households and property investors first, then advisers, brokers and employers who provide it to their clients.
Model
Free plan with manual entry. Paid annual plans with AI document reading (Plus A$79/yr, Portfolio A$149/yr), plus enterprise and white-label options.
Route to market
Direct sign-up through the product site and public demo, then partner channels via financial advisers.

RateScan

Live
Customers
Borrowers, mortgage brokers and lenders.
Model
Free public data and one tracked loan. Pro subscription for multiple loans and competitive analysis (A$35/mo). Bulk and enterprise pricing on request.
Route to market
Organic search through daily briefings and rate pages, plus enterprise enquiries.

AdviceGenie

Prototype
Customers
Advice practices and licensees (AFSL holders).
Model
Per-adviser monthly subscription, with published early-access pricing from A$199 per adviser per month. Licensee deployments by arrangement.
Route to market
Early-access sign-ups and walkthroughs through the adviser site and public demo.

ATS Platform

In use
Customers
Recruitment teams at small and mid-sized organisations.
Model
Hosted platform per organisation. The commercial model isn't settled yet.
Route to market
Not yet defined. The current deployment is the reference implementation.

Current priorities

1 · Grow AssetwiseConvert early-access users to paid plans and release the assistant and adviser portal.
2 · Monetise RateScanBuild the broker and lender offer on top of the free public data.
3 · Take AdviceGenie to pilotBuild the production backend behind the demo with early-access practices.

We also build for others. The engineering behind these products is available to founders and firms as a secondary service: product builds, AI automation and platform management.

View services
Get in touch

Talk to Cognify Labs.

Whether you want to use one of our products, pilot one, or work with us, you'll speak directly with the founder.

CustomersProduct questions, Assetwise or RateScan plans, or an ATS walkthrough.
Pilots & partnersAdvice practices joining AdviceGenie early access, and distribution partners.
Startup ecosystemAccelerators, investors and cloud programmes reviewing the company.
or leave your details and we'll reply

We use your details only to reply. See our privacy policy.