Zeb: How Manufacturers Can Unlock AI’s True Potential
AI promises to transform manufacturing and supply chain operations.
However, many companies are spending heavily and seeing little return.
Mal Vivek, Founder and CEO of Zeb, an AI-focused transformation consultancy, says the reasons are more structural than technological.
"Companies are just having a hard time figuring out how and where to start," Mal explains.
The data problem
Sensors, cameras and automated systems across factory floors and supply chains are generating between 200 and 500% more data than before.
"The problem is that it is so fragmented," says Mal. "It's across so many different systems and it's very heterogeneous – structured, unstructured, semi-structured in some cases."
This fragmentation spans warehouse management systems, transportation management systems and the wider network of vendors and suppliers.
"It's very hard for them to build a cohesive, stitched-together data foundation," Mal continues. "Often, that leads to challenges in getting basic visibility across the entire network."
Proof-of-concept problems
When manufacturers do attempt AI projects, they frequently begin with a proof-of-concept.
“They try to limit the scope by saying, ‘let's just look at this part of the problem’,” Mal goes on. “That proof-of-concept doesn't generate much value because the business looks at it and thinks: you've automated an eighth of the problem, but there are seven eighths still out there.”
The result is a damaging generalisation.
"People do the trial, then label it as AI not being successful," Mal says.
"Once that mentality builds up at leadership level, they lose faith in the technology itself – when actually AI was not the issue. Leadership, process and not understanding the workflow were the problem."
The right people in the room
A core issue, Mal argues, is that AI projects in manufacturing are still being led almost exclusively by technology teams.
"We have to have business leaders in the room," she says.
"We're not just modernising a tech stack anymore. We're really changing the way of working, and that directly impacts the business.
Those business leaders include Chief Supply Chain Officers, Heads of Procurement and those working in integrated business planning and finance.
Mal adds: “Without understanding their process, where the gaps are and how data impacts the way they make decisions, you always have an incomplete picture.”
The way ahead for AI
Common manufacturing use cases, like inventory forecasting, anomaly detection and supply chain control tower management, all share the same underlying requirements.
"It's visibility, context and then action," Mal says. "That's the common thread."
The practical advice is to work backwards from a clearly defined business outcome, identify which data is needed to achieve it, then focus on a specific, narrow workflow – rather than attempting to tackle the entire data landscape at once.
Mal advises: “Take a small, defined workflow as your first use case and start by reimagining that with the context of what AI is capable of.”
Building momentum across the business
This targeted approach, Mal says, creates its own momentum. Once one team sees results, others begin to connect the dots.
Databricks, a data and AI platform that specialises in bringing fragmented data together into a single environment, is one piece of tech used by Zeb to accelerate this process.
Its Unity Catalogue product, which adds security, governance and contextual labelling to unified data, helps create what Mal describes as a shared language across the business.
"Once business leaders start seeing some of the power of it, it sparks a ripple effect," she concludes.
“Democratising AI across the organisation is so critical because without everyone's buy-in, it simply does not work.”
