Hazy - Pitch Deck

Smart synthetic data Preferred synthetic data partner

What is smart synthetic data? Highly realistic test data Smart synthetic data is highly realistic, automatically generated test data. It looks, feels and works just like real data, whilst being safe to work with and fully GDPR/CCPA compliant. “By 2022, 40% of data used to train AI models will be synthetic.” © 2020 Hazy Limited. Private & confidential.

Why does synthetic data matter? Breaking the shackles Reduce risk and costs – reduce attack surface of live data in circulation – reduce risk and cost of compliance Increase speed and agility – eliminate privacy and security concerns – unblock change and innovation “We use Hazy to unlock our data and reduce risk.” © 2020 Hazy Limited. Private & confidential.

Case study In 2018, TSB Bank launched a new system without proper testing because they couldn’t get access to realistic test data in time. As a result, they: – lost 1.3 billion customer records – paid more than £300m in compensation – left 2 million customers without online banking – resulting in the loss of 80,000 customers The CEO was forced to resign. © 2020 Hazy Limited. Private & confidential.

Where is synthetic data used? Data-dependent functions across the enterprise Function Testing Analytics Cloud Provisioning Innovation Proposition Safe, hyper-realistic Safe, statistically Safe data to facilitate Share data across legal Share realistic for test data identical “test” data cloud migration boundaries vendor evaluations Roles QA Manager Data Scientist, CDO, CIO, Data Architect, Innovation Scout Test Manager ML Engineer Cloud Architect Data Provisioning Procurement Benefits Properly test apps and Speed up and unblock Use the cloud without Provision sensitive data Test vendor capabilities services before go live projects and iteration exposing your data quickly and easily using realistic data

Hazy for Testing Your live Catch bugs before they data hit production. Problem Hazy for ✗ risky, slow, poor quality test data Testing ✗ leads to bugs and delays Solution ✓ automatically generate realistic, human-friendly test data Safe, realistic test data ✓ support for modern API and microservice based systems ✓ maintains referential integrity across multiple data sources Benefits ✓ fast, safe, high quality production proving QA / UAT High quality testing © 2020 Hazy Limited. Private & confidential.

Hazy for Data Science Your live Stop waiting for data data Problem ✗ data science and analytics requires access to live data Hazy for ✗ this blocks projects and slows down experimentation Data Science Solution ✓ explore data and train models on smart synthetic data Smart ✓ statistically equivalent, maintains patterns and correlations synthetic data ✓ explicit, quantified utility, similarity and privacy guarantees ✓ supports common machine learning techniques Data Feature Benefits preparation engineering Exploratory ✓ unblock access to data data science Algorithm Model ✓ speed up projects and reduce red tape testing training © 2020 Hazy Limited. Private & confidential.

Hazy for Cloud Migration On-premise Use the cloud without data exposing your data Hazy for Problem Cloud Migration ✗ migrating workloads requires moving data to the cloud ✗ opens up new attack surface and management headache Only the generators move Solution ✓ train synthetic data generators on-premise ✓ move just the safe generator objects to the cloud Hazy Hub and Client ✓ create a synthetic twin of your data platform Benefits ✓ harness cloud resources, APIs, vendors Labs APIs Vendors Cloud resources ✓ without moving any data off premise © 2020 Hazy Limited. Private & confidential.

Hazy for Cross-Silo Train local Analyse data across silos generators on siloed data Problem ✗ data stuck in different silos (legal, geography, department, Sync generators data centre, database system) ✗ can’t merge and analyse to get cross-silo insight Solution Hazy for Cross-Silo ✓ train synthetic data generators at the edge, in each silo ✓ sync generators and aggregate synthetic data, with Aggregate ✓ distributed referential integrity and differential privacy synthetic data Benefits ✓ unlock the ability to analyse data across silos Analysis Cross-silo data analysis ✓ build market leading simulations and predictive models © 2020 Hazy Limited. Private & confidential.

Hazy for Innovation Your live Quickly test and evaluate data ideas and vendors Problem ✗ ideas and vendors often need to be tested on real data Hazy for Innovation ✗ leads to delays, which slows down pace of innovation Solution ✓ provision sandpits and labs with safe synthetic data Smart synthetic data ✓ allows for fast prototyping and validation of ideas ✓ prove vendor capabilities before resourcing live data Benefits ✓ increase speed of innovation Labs Sandpits Test ideas ✓ without risking live data and vendors © 2020 Hazy Limited. Private & confidential.

“Synthetic data technology, of the kind being pioneered by Hazy, is massively important with potential to reshape the data economy and beyond.” Professor Anthony Finkelstein CBE FREng DSc. Chief Scientific Adviser for National Security to the UK Government. “Hazy has the potential to transform the way everyone interacts with Microsoft’s cloud technology” © 2020 Hazy Limited. Private & confidential.

Hazy.com © 2020 Hazy Limited. Private & confidential.