How to build your company’s AI strategy: Last month, we explored how the government has made artificial intelligence a national priority. We spoke with the former ICT Ministry’s advisor for AI Sally Radwan about the government’s national AI strategy and how it is being implemented (Part 1 | Part 2). But despite this interest and increasing adoption by a number of private sector companies, mass integration of AI systems across businesses hasn’t been easy, as companies struggle to figure out how they can deploy the technology.

We put these issues to companies building AI systems, asking them what it takes for companies to build and implement their own AI strategies and departments, as well as needed infrastructure development and possible incentives for private sector companies. Enterprise hosted a roundtable discussion with these players to discuss how companies can best leverage their AI potential. Our guests are:

• Ahmed Abaza, the founder and CEO of Synapse Analytics, which advises other companies on how best to infuse AI into their businesses.
• Ahmed Roshdy, the analytics and data science lead at Vodafone Egypt. He runs the 18-person strong department that’s been focusing on integrating AI into the company’s operations.
• Mohamed Soliman, the data, AI, automation and security leader at IBM Egypt. IBM has been at the forefront of AI adoption and development globally, and Mohamed will give us some insight on what the company has been upto in Egypt.

** LISTEN TO THE DISCUSSION AS A PODCAST (runtime: 44:01 on our website) or tune in via Apple Podcasts | Anghami | Google Podcasts | Spotify). Or you can read edited excerpts of our conversation below:

AI adoption in Egypt is growing, especially since covid. “There is no doubt that the covid pandemic had a big role to play in accelerating all types of digital transformation, including AI adoption,” IBM Egypt’s Soliman told us.

There are a number of sectors here currently focusing on AI integration and innovation: Besides pure tech companies like IBM and Oracle, telecommunications and banking are heavily looking into AI strategies, our guests told us. At Vodafone, we’ve been building an engine that recommends promotional offers to our customers, Vodafone Egypt’s Roshdy said. “Currently, we have more than 11 mn customers that are gaining offers from that platform,” he added. With the rise of fintech and the vast spread of telecommunication services, companies are looking at how to best optimize processes and come up with innovative solutions to serve their customers.

Companies with large amounts of data appear to be the most interested in AI, Synapse Analytics’ Abaza tells us. Hence, retail and e-commerce are also high on the list.

But even key national projects are seeing AI adoption: “IBM was awarded a project with the Egyptian Holding Company for Silos and Storage (EHCSS) that would see it automate the entire wheat [storage] cycle,” said Soliman. This includes the entire supply chain, from sensors in the silos that accurately measure how much wheat is stored, to measuring temperature and humidity in those silos.

So, what’s been getting in the way of adoption? Issues with data: “The main challenge to adopt AI varies from one organization to the other, but sometimes it’s the data complexity and the fact that data exists in silos,” Soliman explained. Abaza agrees, adding that the argument about Egypt not having enough data does not ring true. Data does exist across companies, but it exists in silos. Oftentimes, enterprises have a hard time leveraging that data and actually use and understand it efficiently.

Another major internal challenge to companies are apprehensions about the impacts of AI to an organization. There are fears of employees getting replaced by machines, which could curb internal AI adoption in companies, Abaza said. While Roshdy confirms that employees are prone to lose their jobs to AI in some instances, this is a chance for them to seek more advanced jobs. Egypt’s labor market is very stacked and a lot of jobs can be very easily replaced, Abaza said. “We still have people that push buttons on elevators.” But Soliman reminds us that in the end, the job of AI is to augment humans, not replace them.

And some work needs to be done on the infrastructure level: “This is something that doesn’t end because you keep on improving what you already have,” Abaza explained. Some improvements are needed when it comes to internet infrastructure, specifically, because high-speed internet connection is necessary to test use cases, he added. Moreover, there is no local, mature software to manage and scale AI, he said.

But there are incentives for private sector companies to start experimenting with AI: “The government is investing a lot in AI capacity building. That’s one piece of the puzzle, Soliman said. Additionally, when the government’s Center of Excellence AI lab is launched, it will help companies develop use cases with government support, he added. The lab is essentially a platform where practical tools are developed and then fished over to the private sector to apply them, ICT Ministry adviser Sally Radwan told us.

So how can a company get started with its AI department? There are several ways: Abaza suggests hiring a business analyst, data scientist, software engineer, and machine learning operations engineer to get the department started. “The starting point is hiring a lead to fill the gap between technology, data science and business,” Roshdy recommends. That lead can then hire the team mentioned by Abaza. Soliman also suggests leveraging software tools that are built to help companies get started, instead of hiring everyone at once. The key is starting with a small pilot project before going full Skynet.

The main focus should be to collect data in a meaningful way: AI without proper, quality data is useless, Soliman tells us, adding that the hard part lies in data collection, organization and quality. Larger enterprises that have been around for 50 or 60 years face a challenge in making sense of the data they’ve collected over the years as it isn’t organized and wasn’t collected with a certain goal in mind. Abaza agreed, saying that some companies do not know how to use their data or how qualified it is. “Very valuable data, the crown jewel of any organization, can be used to create very valuable insights,” Soliman explained.

But make sure you have a foundation in place that protects the data privacy of your customer. “Every use case […] should be aligned with an internal privacy officer,” Roshdy said. That person, along with his team, should set guidelines on what information the company is collecting and how it is being shared internally to ensure that no privacy is violated. And the whole company needs to be aligned with that policy. This could restrict the usage of data for some analytics, but it ensures that the company remains compliant to data privacy laws. The customer information should be encrypted in all cases, while the privacy officer needs to approve every use case, he added.

And with great data comes great responsibility, but that shouldn’t stifle innovation: Abaza commends how encompassing the Data Privacy Act was. But while privacy is essential, innovation should not be curbed in the process, all our guests emphasized.

---

Your top stories on future trends for the week:

• Egyptian healthtech startup Doxx raised USD 1.5 mn in a seed round led by international VC outfit Openner, with participation from Egypt’s Elevate Private Equity.
• Cairo-born Instabug closes USD 46 mn series B round: The round was led by New York-based software investor Insight Partners, with participation from existing investor Accel, as well as new investors US-based VC Forgepoint Capital, and Endeavor.
• E-commerce startup Teegara has successfully closed a five-figure USD bridge round led by Alexandria Angels Network, with participation from new angel investors in Egypt, Saudi Arabia, Jordan and the Netherlands.