Listen or watch now on
YouTube, Spotify, or Apple Podcasts
For Anis Harb, the real barrier to enterprise AI adoption is not access to better models. It is everything that sits between a promising demo and a system that actually works inside a business.
After nearly a decade at Deliveroo, where he launched and led the company’s Middle East operation, Anis left with a growing frustration: as businesses scaled, they added more headcount, which created more complexity, more process and, eventually, the need for even more headcount.
Algebra AI is his attempt to break that cycle.
The Dubai-based company emerged from stealth in June with $7M in funding, backed by Infinity Constellation, BECO Capital, Silicon Badia and Waseel Investments.
Founded in partnership with that investor group, Algebra is targeting the more than 30,000 mid-market businesses across the GCC that sit between off-the-shelf AI software and the cost and complexity of enterprise-grade deployments.
Rather than selling businesses another piece of software or building a workflow and walking away, Algebra operates as a managed AI service. Its team works across the data layer, workflow logic, integrations, edge cases and human approvals, then stays embedded to monitor, maintain and improve the system over time.
As Anis puts it, Algebra wants to be the accountable operator, responsible not only for getting an AI system live, but for ensuring it continues to run as intended and delivers measurable business results.
The company already has engagements across financial services, food and beverage, manufacturing and distribution. While the industries may look broad, Algebra is deliberately narrow in the repeatable workflow patterns it takes on, identifying high-ROI processes that can be transformed from hours of manual work into minutes, or help businesses make decisions and take action in real time.
We also get into Anis’s unusual journey into Deliveroo, why he originally planned to build his own food-delivery company in the Middle East, how AI is changing the structure of Algebra’s own team, and what his decade inside one of the region’s most important consumer platforms taught him about localisation, growth and operational complexity.
A massive thanks again to Anis for taking the time to come on the FWDstart podcast.
We cover:
Why enterprise AI projects so often stall after the initial pilot, and why workflow logic, data infrastructure and ongoing ownership matter more than the demo
Algebra’s managed-service model, from identifying a high-ROI use case to staying embedded as the accountable operator responsible for its performance
How Anis’s decade at Deliveroo shaped his belief that companies should be able to grow without automatically adding more people, process and complexity
Why Algebra works across multiple industries but stays disciplined around repeatable workflow patterns rather than taking on entirely bespoke problems
How AI is changing team design and productivity, including Anis’s belief that five AI-native engineers can now achieve what once required 20
Why global AI-services playbooks cannot simply be transplanted into the GCC, where labour economics, regulation, data sovereignty and local operating behaviour require a different model
Timestamps
00:00 - Why AI Adoption Breaks Inside Businesses
00:49 - Leaving Deliveroo After a Decade
04:21 - The Startup That Nearly Became Deliveroo Middle East
06:50 - Why Managed AI Services Are Booming
08:17 - Turning Hours of Manual Work Into Minutes
10:06 - What Makes a Business Ready for AI
11:50 - Why Algebra Stays Embedded After Launch
16:33 - How Do You Price Managed AI?
18:46 - Taking the Palantir Model to the Mid-Market
20:42 - Why AI Services Must Be Localised for MENA
22:19 - Building an AI-Native Company With Fewer People
31:08 - Deliveroo, Food Delivery Consolidation, and the Fight for Market Share
Subscribe to the show
I’d love it if you’d subscribe and share the show. Your support makes all the difference as we try to bring more curious minds into the conversation.
For inquiries about sponsoring the podcast, email [email protected].
