Nov 26,2020

The AI Journey: How to deliver success

By David Gristwood, Application Architect – Microsoft UK

Over the past five years, I’ve been involved in dozens of hackathons. These events, which bring together architects, developers, and programmers, for a design sprint, have helped hundreds of partners take their ideas and turn them into code and working solutions.

Hacks are a great environment for technical teams to get away from the bustle of their day to day work and – with the help of Microsoft subject matter experts – explore ways in which they can design, test, and build new software and systems. Over the last year or so, more and more of the projects have centred on AI, as partners look for ways to add AI capabilities to existing software or build new solutions or services with AI at their core.

However, the path to success and deliverable solutions isn’t always an easy one. I thought it useful to share some common issues I’ve seen, and how you can overcome those challenges.

1 – Getting the right data and tools

For many software companies, a background in more traditional web and database systems isn’t a great grounding for AI. Many struggle to separate the hype from reality when it comes to this emerging technology, and to make the correct choice about what tools and technologies to use.

When joining a hack with the goal of building the best possible machine learning models, it’s important to bring good quality data.

AI is only as good as the data it’s fed. Poor quality data churns out poor quality results; no data offers no results. To ensure a successful AI journey, identify what you want the technology to do, then gather as much related data as possible for the AI systems.

Best practice: Understand what you want AI to do, as this informs what data you need and what tools you’ll need to implement it.   

2 – Spend more time on the exploration stage

To successfully implement AI, a business should spend considerable time assessing how AI can help them. This exploratory stage is critical, as it allows you to define the scope, scale, and purpose of any AI project.

We’ve learnt that time is best spent on defining realistic goals and ideas with a clear view of the business benefits AI will bring them.

Without taking the time to define objectives and explore why AI is right for the business, progress is likely to be slow.

Best practice: Take your time during the initial phase of any AI plan. You should have a clear idea of what you want AI to improve, and an understanding how it fits into the larger business goals.

3 – When to bring in the experts

With so much promise, it’s easy to get caught up in the AI hype. Buy into it and suddenly, science-fiction becomes science-fact. Some organisations are so swept away that even armed with a distinctive business plan and a great team beavering away on the data, they jump right in without any real knowledge of how to implement the technology. Such enthusiasm is to be admired, but it can be reckless and costly.

Most organisations are new to the field of AI, even if the company has extensive IT experience. In this situation, every partner should look to team up with those who are well-versed in AI. Microsoft has plenty of subject matter experts on hand, ready to help guide you through the AI process.

Best practice: Explore Microsoft’s AI learning tools available online, from learning how to create your first AI app to the popular AI business school that looks at strategy, responsibility, and the technology itself – and everything in between.

Join the journey

When we noticed those trends, we realised that to take partners from AI inception to production; from experimentation to implementation, we had to have a clear journey.

This led us to the ‘AI journey’ – a repeatable, scalable series that would get as many partners skilled, as quickly as possible.

This journey takes the form of events and activities. The goal is to skill up partners just like you, giving you the opportunity to improve your skills and knowledge through a structured, step-by-step program. You’re also free to drop in and out, joining the sessions that match your experience level and stated goals. 

Key steps in the programme include…

  • Understanding AI, and showcasing scenarios in which AI added real value
  • Exploring Azure Cognitive Services and how to use these prebuilt AI building blocks, covering areas such as speech, text and vision
  • Getting to grips with the basic concepts behind machine learning and algorithms
  • Demonstrate the tools and technologies available to productionise and deploy these machine learning models
  • Delve into deep learning AI areas and more advanced data science techniques

Getting results

Partners who have already been in this journey find themselves better prepared, better skilled, and more successful. For example:

A car telemetry software company was able to help insurers check if drivers had actually broken the speed limit – a previously slow and error prone task.

A marketing company that produced expensive, high-quality brochures now use AI algorithms to predict likely customers, dramatically cutting sales costs.

An industrial printing company saves £50k a year on printer ink spoilage costs in the factory by using AI to control cartridge refills. This was the result of spending just three days at one of our hacks, focusing on just one step in the manufacturing process.

There are lots of resources to help partners on their AI journey, from inspiring them to innovate to detailed technical articles, labs, and videos. We’re also running our AI journey training events in London between October and December, and registrations are now open at the AI Learning Journey hub.