Stolen authentication data is the cause of more than a quarter of company security breach cases. That is why, as Deloitte experts predict, the understanding of online security will change over the next two years.

Three of the nine technology trends developed in the latest report by the consulting firm "Tech Trends 2021" are the spread of the "zero trust" philosophy, the MLOps methodology and data management in such a way that machines can use it.

The selection of the latest technology trends was determined by information obtained from the management boards of the companies we serve regarding current and future priorities, forecasts and analyses prepared by scientists, and finally the investment plans and priorities of leading startups, venture capital funds and technology providers. The information gathered by Deloitte experts shows how unexpectedly the COVID-19 crisis forced changes. More and more organisations across all sectors are accelerating their digital transformation process, not only to increase the speed and efficiency of their operations, but also to respond to rapid fluctuations in demand and customer expectations – says Daniel Martyniuk, partner, leader of the IT strategy and transformation team at Deloitte.

From homemade methods to engineering precision

From homemade methods to engineering precision

As machine learning (ML) and artificial intelligence (AI) become key factors influencing organisational performance, business leaders are beginning to realise the need to accelerate the implementation of these solutions.

Typically, these models are created and deployed using manual, custom, difficult-to-scale processes that prevent experimentation and hinder collaboration between product teams, the IT department and data analysis specialists. In other words, the enterprise infrastructure is not designed in a way that enables fast, consistent and efficient creation of machine learning models.

The spread of the MLOps, or Machine Learning Operations, methodology is an important topic of the latest "Tech Trends" report. Our observations show that companies will increasingly move from "homemade" implementation of artificial intelligence and machine learning solutions to an automated and systematised methodology that will allow IT, data science and operations teams to implement and monitor such solutions more effectively. This will give organisations the ability to react quickly and make accurate decisions – Agnieszka Zielińska, partner in the Financial Advisory Department, Deloitte.

To integrate AI and ML with all processes and systems in an organisation, companies must be able to implement them in a standardised way and on a large scale. Twenty years ago, similar challenges led to the birth of DevOps – a software development method based on communication and collaboration between developers and operations specialists.

Through standardisation and automation of application creation, deployment and management, DevOps changed the way many IT teams build and deploy software, enabling them to improve its performance and quality.

MLOps is an approach that combines and automates the development and maintenance of machine learning models, aiming to accelerate the entire lifecycle process. By accelerating the experimentation and model creation stage, facilitating their monitoring and managing regulatory requirements, MLOps helps companies increase business value. The value of the MLOps market is expected to grow to nearly $4 billion by 2025.

The data revolution

The data revolution

It turns out that traditional ways of organising data are not enough to make decisions based on artificial intelligence.

A growing number of companies that pioneer the use of artificial intelligence are noticing the incompatibility of data structures and models with new technologies focused on machine decision-making. Until now, these structures have been tailored to a typically human way of thinking. We expect that more and more enterprises will begin to actively face this challenge – says Aneta Olędzka, manager in the analytics and cognitive science team, Deloitte.

Organised data as the foundation of effective decisions

Organised data as the foundation of effective decisions

One of the trends we will observe over the next two years is the development of a new approach to data management for use by machine learning algorithms rather than by humans.

People tend to look at aggregated data characterised by two or three main factors. When faced with more complex data, many have difficulty processing the information presented and making decisions. Machine learning models can extract low levels of statistical significance from huge amounts of structured and unstructured data.

In Deloitte's State of AI in the Enterprise study, when respondents were asked about the most important initiative to increase their competitive advantage through artificial intelligence, they answered "modernising our data infrastructure for AI". For companies that are at an early stage of digitalisation, the stakes are particularly high. Digitally advanced companies, often unburdened by technical debt, with new data models and processing capabilities, begin to reap financial benefits faster thanks to them.

Zero trust, or security differently

Zero trust, or security differently

Conventional cybersecurity models are proving insufficient to meet constantly changing cyber threats, especially in the face of constantly evolving business models and a dynamically changing workforce. With the growing number of cloud-based systems, remote workers and the multitude of devices connected to a company's network, security boundaries are constantly shifting. The predicted development of smart devices, 5G, edge computing and artificial intelligence heralds an additional expansion of the potential attack surface.

This is not another episode of "The X-Files". "Zero trust" is a philosophy that changes the approach to understanding what network security is. We know that most networks are difficult to break from the outside, while from the inside access to most resources is practically unlimited. In the latest "Tech Trends" report, we point out that there is a specific set of both security measures and activities or concepts that should be implemented to properly secure an organisation – says Adam Rafajeński, director in the cybersecurity team, Deloitte.

External security measures emphasise the authentication of users and devices connected to the organisation's network. That is why the theft of authentication data is the cause of more than a quarter of security breach cases. Meanwhile, a properly designed "zero trust" architecture is the basis for effective control and management of user access, while reducing operating costs and supporting scaling to tens of thousands of users. Similarly, onboarding employees, contractors, cloud service providers and other suppliers can become more efficient, flexible and secure.

Carefully designed "zero trust" architectures can interoperate with other automated IT practices, such as DevSecOps and NoOps. A key element of this philosophy is the microsegmentation of networks, data, applications, workloads and other resources into individual units to limit breaches and ensure security control at the lowest possible level. By limiting access based on the principle of least privilege, a minimum number of users have access to data and applications.

By removing the assumption of trust from the security architecture and introducing authentication of every action, user and device, the "zero trust" principle helps create a more robust and resilient security foundation. The organisational benefits are complemented by a significant advantage for end users: seamless access to the tools and data necessary for efficient work.

Summary

Deloitte's "Tech Trends 2021" report identifies three key technology trends: the "zero trust" philosophy, the MLOps methodology and data management for use by machines. Experts emphasise that the COVID-19 pandemic has forced the acceleration of digital transformation, and companies must face new challenges, such as the theft of authentication data, which accounts for more than 25% of security breaches.

MLOps, or the automated deployment of machine learning models, has a chance to become a standard similar to DevOps in software development. Meanwhile, the "zero trust" architecture assumes a complete change in the approach to network security — instead of relying on external security measures, organisations should authenticate every action and user. By 2025, the value of the MLOps market is to grow to nearly $4 billion.