The AI System: The future of the enterprise

To put it into context, in the same way that business functions demand their own ‘System of Record’ – sales and marketing have CRM, operations have ERP – AI is now so fundamental to the enterprise that it demands a new system. We call this the enterprise AI System. 

The AI System will power decision making at every stage of a business’ value chain. It won’t be constrained functionally like current workflow systems, and it will consider the global optimization of a business as its primary goal.

Without this new system, harnessing the true power of AI will simply not be possible. The current software ecosystem means that there’s a distinct lack of scalable AI infrastructure, which creates dispersed data across the enterprise. There’s also nowhere to deploy AI in an end-to-end workflow, which means the process of building AI solutions becomes laborious and slow. On top of all this, there are insufficient integrations and support services – which means managing AI’s success is just about impossible.

A new way of looking at enterprise architecture is required, alongside new technology to power that new way of thinking.

Getting to know the AI System

For any business leader looking to supercharge their enterprise using AI, there are some fundamentals that you simply have to establish, fast. You have to eradicate data silos, bridge the gap between disparate systems, put intelligence at the very core of the business, and actively push it to deliver actions and outcomes.

Each of these core pillars will serve value to the organization in more ways than one. Think about it; what would you do with all of your data in one place? What if you knew how a change in marketing spend would impact your supply chain? What if you could rapidly test, learn and improve your decision making? The intelligent combination of all of these underpins the AI System, which further multiplies the value it offers an enterprise.

No silos

Peter Thiel describes data network effects brilliantly in his book, Zero to One. To paraphrase; companies that harness data to gain a competitive advantage create a flywheel effect, where their ability to service customers is enhanced. This leads to them winning more customers, generating more and more data as they grow, which, in turn, reinforces their advantage. This data network effect can be used to capture the majority of a market’s economics (profit) and see companies go from – you guessed it – zero to one.

This is why it’s absolutely crucial that you eliminate all of the data silos in your business.

AI has the unique ability to consume data on a scale that we’ve simply never seen before. In order for it to excel in its decision making, AI needs to access as much data as possible. Walled gardens of information have been created in order for us mere humans to be able to consume the data in manageable, bitesize chunks. This means that decisions are made – decisions which affect the whole business – often without the right information being taken into account. More often than not, this leads to unintended consequences and less-than-optimal outcomes.

The reason I use the Thiel example is that exposing AI to as much (useful) information as possible allows it to understand the game you are playing in, as a business. The more data you provide AI with, the better it becomes at recognizing patterns and predicting outcomes. In the same way that humans learn from experience, you could argue that AI behaves in a similar way.

Innovative companies like Dremio have recognized this as a ‘must do’ for businesses who want to win in the AI era. Their ‘Data Lake Engine’ abstracts away the need for warehouses, cubes, aggregation tables and extracts. The output from ‘engines’ like this will feed directly into an AI System, creating immediate opportunities to drive real business outcomes.

By gathering more data, and by presenting that data in the right way, you are giving the AI an advantage to make more informed decisions. The more informed decisions you make, the more data the AI can learn from, and so on and so forth. It’s a data network effect – but it all starts with eliminating the data silos in your business.

Centralized intelligence

At some point during our careers we have sat down to an Excel spreadsheet with more rows and columns than grains of sand on the earth. We have saved that spreadsheet, emailed it, copied it, adapted it and only then (plus a few more circles of that chain) made a decision on it. It took months to build, another month to gather the feedback and another month to make those decisions a reality.

Why? Because we had no access to a central intelligence. As businesspeople, we created a system that got the information needed to make a decision into the hands of the people as quickly as possible. The problem was – it was still painfully slow!

Centralized intelligence flips this concept on its head. Rather than passing the ‘intelligence’ from team to team over the course of many months, we need to bring the teams to the intelligence. You need to create a decision making and information hub that all teams can use, simultaneously – exposing immediate clarity, pace of action and visibility of outcomes across the organization.

The problem at the moment is that many software solutions, from customer email to warehouse management systems, are created to follow this ‘conventional’ human workflow – a workflow that we just identified as being oh-so problematic to rapid growth.

AI changes this relationship with software and systems. It allows all of these diverse systems to communicate, and it allows all of these systems to become a key input into centralized intelligence. But, this centralized intelligence has to reside somewhere, in a new system – an AI System.

We’re working with businesses who were stuck in the old way, but, by shifting to a centralized intelligence model, we’re helping them do great things. Take our customer, boohoo, for example. Within the last 12 months, the fast fashion giant has moved to a centralized intelligence powering core functional teams. They just did this.

Actions and outcomes

As businesses we live and die by our P&L, and as business leaders we live and die by our decisions that affect that. It’s our responsibility to take actions and drive outcomes that positively influence this. We empower our teams to do this, and good management is about how effectively we make this happen. Why then, as technology leaders, do we allow 80% of data projects, according to Gartner, to fail? The answer – because they are not focused on actions and outcomes.

By its very nature, AI is focused on outcomes. It’s built to solve problems and to answer the ‘why’, but, more importantly, drive the ‘so what?’

It learns by understanding its decisions, making improvements on them and then iterating that process. For that reason it is paramount to ensure that AI can power as many decision making processes as possible.

As it stands, the enterprise ecosystem makes it very difficult to put AI solutions into your workflow systems and drive outcomes. The technology was created at a time when this was not a reasonable demand of its functionality, but now it is.

Footasylum, a UK streetwear retailer, was an early adopter of this actions and outcomes first approach. In using the Peak AI System, the business drove an increase in digital marketing revenue of 28% YoY. But that was just the start. Footasylum quickly realized that this centralized AI was capable of powering its acquisition strategy, too. In integrating with its acquisition platform, Footasylum was able to increase its average ROAS by three times the current level.

As the Footasylum story shows, you need to integrate and push the decisions made by the centralized intelligence into these systems of engagement in order to benefit from your AI-driven decision making. Your AI System has to be agnostic to integrations; able to connect seamlessly with any other system in your organization.

Businesses like Tray.io have identified that this ‘glue’ is going to hold these systems of engagement together with their centralized AI Systems, by creating a centralized automation platform. Gone are the days of buying tin and wire for your basement – these are the decisions that technology leaders are taking to ensure they are ready to drive outcomes in the AI era.

Summary

To summarize, it’s clear that, as businesses, we must all adapt if we are to grow and win in the AI era. This will mean a new way of thinking about our enterprise systems and architecture. The exciting part, we believe, is that these new systems present us with an opportunity to create growth on a scale that we haven’t experienced before.

We are already seeing AI-powered businesses take advantage of the transformational benefits the technology offers them (UK retailers are using AI to grow 30% faster than their peers, with 50% higher profit margins). There is no question that we stand on the edge of the AI-era. The time to act, for those not in the game already, is now. 

More from Peak…

? We’re at the dawn of the AI era. Find out why in this note from CEO Richard Potter

? Peak secures $12M extended series A investment to power AI System

? Peak secures its spot in The Sunday Times Top 100 Small Companies to Work For

? We sat down with Intel for an AI fireside chat