One platform to AI it all: Google Vertex AI makes AI/ML simple in SA


Johannesburg, 03 May 2022

South African enterprises are increasingly looking to adopt artificial intelligence (AI) and machine learning (ML) to innovate, improve processes and enhance customer service, but many are finding their efforts hampered by a lack of skills and the complexity of the AI/ML environment.

This is according to Louis van Schalkwyk, Head of Technical Operations at Digicloud Africa, who says AI/ML offers compelling benefits, with sectors as diverse as retail and finance through to agriculture looking to harness it. “Many companies simply don’t know how to get started, or to advance their AI/ML roadmap,” Van Schalkwyk says.

“To benefit from AI/ML, organisations need more than just the right technology,” he says. “To build an effective AI capability, they must also overcome challenges around the people who use the technology, the data that fuels the models, and the processes that govern all of it. A key challenge with AI/ML is ensuring that these four areas and their interplay with each other is managed in a way that ultimately leads to a transformational maturity that provides organisations with increased market share/competitive advantage. Just focusing on one or two areas alone isn’t enough; it’s about effectively managing all the different aspects.”

Van Schalkwyk explains that companies just starting out on their journey can tap into Google’s pre-trained ML products to take advantage of tools for sentiment analysis, image or video analysis or converting text to speech (or speech to text) and translating between languages. “These pre-trained ML models are ready to use and don't require any custom models or even your own data, you just call the API and get your result, so it's really simple and quick to integrate into existing applications,” he says.

Companies that are looking at using ML to gain a competitive advantage in the market should focus on multiple custom models that are continuously evaluated, retrained and updated in an automated fashion, he says. “For these organisations, Google’s Vertex AI offers a single platform to do it all.”

Vertex AI is a single place (UI and API) where you can train, compare and operationalise your ML models. It provides a centralised repository where your models can be stored and shared between different teams in the organisation.

Says Van Schalkwyk: “For companies that aren’t ready to build their own custom models just yet, Google offers AutoML as part of VertexAI. AutoML allows you to create high quality custom ML models from your data, whether it’s image data, tabular, text or even video.

Vertex AI Pipelines provides all the tools and interfaces required to apply DevOps strategies to your ML models – MLOps. The service helps you to automate, monitor and govern your ML systems from start to finish in a serverless manner. Artefacts are stored in Vertex ML Metadata, so that you can analyse the lineage of your workflow’s artefacts.”

Google Cloud’s framework for AI adoption

Google Cloud’s framework for AI adoption provides a guide to technology leaders who want to build an effective AI capability, enabling them to leverage the power of AI to enhance and streamline their business, smoothly and smartly.

Google’s framework for AI adoption is based on four key areas – people, process, technology and data.

The interplay between these four areas give rise to six themes:

  • Lead – how teams are structured, level of executive sponsorship for ML projects, how budgets are allocated to ML.
  • Learn – how to gain, develop and retain ML talent.
  • Access – looking at processes for the creation and sharing of datasets and ML assets.
  • Scale – allocating capacity for ML workloads, provisioning of cloud resources and using specialised hardware for faster training.
  • Secure – establishing trust in your AI, looking at security controls to protect data and the ability to explain decisions made by your AI.
  • Automate – how do you keep your models current and updated through continuous training and deployments.

Google’s framework deep dives into each theme to help an organisation understand where they are in their AI journey (tactical, strategic or transformational) and what to focus on in each theme in order to progress.

Van Schalkwyk says: “It’s a framework that’s technology and platform agnostic. It’s aimed at IT leaders, at start-ups and enterprises to help them map out a successful roadmap to building out an effective AI capability.”

Digicloud is Google’s enablement partner for Google resellers in Africa. Through a network of 70+ resellers across Africa, Digicloud is able to help customers adopt and develop effective AI capabilities by matching them with the right reseller based on their business needs.

Share

Editorial contacts

Lauren Wood
Marketing and Onboarding Lead
(083) 276 1932
lauren@digicloud.africa