Main content

A mix of AI and processes is the real key to getting value from data, says Celonis

Madeline Bennett Profile picture for user Madeline Bennett June 28, 2024
Summary:
Process mining firm Celonis advises customers that true ROI comes after 18 months

An image of the Celonis event in London and speakers on stage
(Image taken by author)

Following hot on the heels of a raft of AI updates earlier in June, Celonis hosted its user event in London last week (18 June), where it fleshed out its proposition to be both the Wikipedia of enterprise AI, and a guardrail against issues like hallucinations. 

Rupal Keria, VP & Leader for UK&I at Celonis, used his keynote speech to reiterate Celonis’s core premise: processes are the greatest lever for value and the fastest lever for change in any organization. He noted:

Improving cash, improving productivity, improving cash flow, profit, whatever it may be. You do that by focusing on your processes first. Not on your IT systems, not on anything else. If you understand your processes, you can then focus on everything else.

Karia shared some examples as evidence of this, including one of its supply-chain clients, where one in five orders was not complete due to stockouts, despite the items being in stock. 

You've upset a customer, you've then not got your top line growth because you've sacrificed 20% of your orders potentially, and you've also impacted your bottom line. By doing some minor tweaks to processes, we could identify that.

But while the benefit of taking a process approach is clear at Celonis, what's less obvious is why uptake is not moving at the momentum and pace the firm would expect. Karia cited a disconnect in enterprises that's slowing down the ability to improve processes, and hence improve business performance. Karia said:

If you are in a world of supply chain, you think about SKUs or shortages; if you're in finance, you're thinking about ROI; if you're IT, you're thinking about configuration and tables. The challenge is, these systems don't work well together, but each one of these systems give you hugely valuable data.

Unless firms can connect those bits of information together, it's meaningless, Karia adds. That’s where Celonis comes in, offering what it calls the ‘connective tissue’. 

What we are able to do is understand at each level, regardless which part of the organization, what's going on, pull that together to a common language. Once you've got that common language, you can work your processes out regardless of whether you're talking about any one of these systems and that's where you get the rich information and the data doesn't lie. But the only way of doing this is getting a common language and a common tissue.

Rather than trying to compete with the big-name enterprise vendors like SAP, Oracle and Workday, Celonis is aiming to capitalize on the data those systems collect, adding this connectivity layer and shared language on top of them. 

We are not sitting here suggesting you replace your IT platforms or your data platforms. We sit on the top of those, and we enhance and improve the ROI of your systems that you have in place today because we are able to understand those systems and how they work to get better productivity from those systems.

Once that information is available, Celonis can go on to obtain a common average and base level of knowledge of those systems and data, which then feeds into a company’s analytics, automation or AI platforms to make those more effective. 

Enterprise context 

Regarding Celonis’ AI offerings, Karia referenced the AI hype, quipping that everyone has built AI into their platform and it's what everyone would expect him to say. But he added:

What's a bit different here is the way we're talking about powering AI.

While AI has certainly taken off in the consumer space, in the enterprise, it hasn't taken off as everyone expected, Karia said. 

That's because the internet and Wikipedia are great at powering ChatGPT at home to design your poems or whatever fun things you do with it. However, that doesn't exist in the enterprise. What we believe we can be, because of the way we collect data, is Wikipedia for the enterprise.

This mirrors previous comments by Divya Krishnan, VP of Product Marketing at Celonis, that what is currently missing from AI in the enterprise is the context of how an organization runs, to fill in the gaps around the data and provide highly meaningful knowledge.

As well as adding this context around the data, Celonis also promises to help prevent AI glitches like hallucinations. Karia cited companies that have been taken to court due to a chatbot or AI algorithm deviating from a process, and losing.  

The reason they've lost those cases is because if you've got an AI model or chat bot or any other type of IT system, that represents you as a company. You have a process. A hallucination is a deviation from a process because it's self-learning to do something it's not supposed to do.

Where AI starts to go askew and deviate from the process, Celonis can identify that and take action. Karia said:

We can stop you and give you that governance and guardrails to make sure you can power AI to a different level.

During a demo of a newly launched AI tool, Celonis showed how its Process Copilot identifies and surfaces the most urgent and valuable opportunities for businesses, by applying AI on top of its Process Intelligence Graph digital twin. The objective is to let anyone in the team harness the process data within company systems and interact with it in natural language. 

Using an example around days payable outstanding (DPO), with issues caused by payment terms mismatches, the user asks for a table showing vendors and materials ranked by DPO. Celonis searches the data and provides a table, showing the top culprits are all in the EMEA region. The user then asks for the trend over time of the EMEA DPO and identifies a steady drop over the past year.

The user is then able to ask – ‘what are the main opportunities for me to fix this?’, and Celonis will give recommendations, in this case using its AI-powered Payment Checker app to match vendor master data, purchase orders and invoices. 

While Celonis is keen for firms to switch over to its process-based approach and reap the rewards of these AI additions, Keria noted that getting ROI from the technology isn’t necessarily a quick process. Working with customers, partners and the Celonis team, the firm has observed it takes around 18 months over three stages to get sustainable, long-term value. 

Phase one covers the first six months, aiming for simple processes and low-hanging fruit where companies can get quick wins. Phase two, which is the next 12 months, sees firms focusing on three to five of those early processes, connecting them together to get rich information. 

Once you’re past the 18-month point, with a tech sponsor, proof points and engaged users on board, firms can move to much more complex processes and hunting for more ways to transform. Karia added:

People assume - we're not a really large IT system, you go in day one, you implement it, you get the value straight away. If you try to jump from being an early adopter straight to doing lots, is it sticky enough? Do you have the buy-in? Do you have the customer change process on board? That's the way we think you scale.

Loading
A grey colored placeholder image