Algebra's Anis Harb on Why Hiring Is Not the Answer to Scale
The co-founder and CEO of the UAE AI transformation company traces his approach to a decade at Deliveroo and an earlier stretch in banking.
Algebra builds and operates custom AI workflows for mid-market companies in the Gulf. It launched in June 2026 with US$7 million from a group of regional and international backers: US-based Infinity Constellation, UAE-based BECO Capital, Jordan-based Silicon Badia and Saudi Arabia-based Waseel Investments. The company works across financial services, manufacturing and distribution, and it keeps deliberately to workflow models it can repeat rather than one-off builds that grow harder to maintain as they multiply.
Anis Harb, the co-founder and CEO, came to the problem sideways. He started in banking at J.P. Morgan, then spent a decade launching and running Deliveroo's Middle East business.
What he carried out of both was a frustration with the way companies absorb growth. Volume rose, and the answer was always more people and more process, until the work of keeping pace became the work itself.
“Data arrived from different teams and markets at different times, and by the time we pieced it together, the moment to act had often passed,” Harb says. “The business had the intelligence it needed. Getting it to the right person in time was the hard part, and that gap is where growth opportunities get lost.”
Algebra's answer is to put AI on the coordination layer, the connective work between systems and teams, so the state of the operation is legible while there is still time to act on it.
“That makes a business more efficient, and it also helps it grow, because the team can respond faster, take on more volume and act on opportunities while they are still live,” Harb says.
The market it targets is specific. The GCC holds more than 30,000 medium-sized companies, a band that tends to be too large for off-the-shelf software and too small for enterprise programmes. Harb puts it plainly: “I could also see who needed it most.”
He had watched the problem up close in grocery. When Deliveroo moved into the category, each new grocer arrived with 10,000 to 20,000 products, and the data behind them came in whatever format suppliers happened to use: spreadsheets, PDFs, photographs of printed catalogues.
“Bringing each new grocer live took weeks of careful work, and keeping stock accurate as it changed every day took ongoing effort from our operations and support teams,” he says.
Adding staff was the obvious fix, and it worked. It also showed him how much of that repetitive load a well-designed system could carry. He hears the same thing from operators now. A contracting COO told him his pipeline is capped by how many tenders his team can price by hand. “That is a revenue problem as much as an efficiency one,” Harb says. Once the constraint lifts, the question shifts from keeping up to growing.
The raw material is familiar. Business data sits scattered across ERP and CRM platforms, email, spreadsheets, documents and WhatsApp threads. People spend their days moving information between those places, checking procedure and chasing sign-off, and decisions land after the moment has gone.
“General-purpose AI can answer questions. Closing the gap takes a system that knows how your business runs,” Harb says. “Every company has its own logic: which customers get priority stock, who signs off on a discount above a certain level, which supplier confirms orders on WhatsApp. That logic is what turns data into the right next step.”
Algebra starts by mapping the operation as it actually runs, handoffs, approvals and exceptions included. Only then does it design a system around that map, wired into the tools the client already uses. The system can take the next step itself: draft a quote using the client's own pricing rules, route an approval, update a record, flag an exception. People keep the calls that need judgement, with the information already in front of them.
Discovery takes one to two weeks. Design, development, testing and integration run six to twelve weeks, with a target of going live in roughly 90 days. Clients keep their existing teams but must supply an executive sponsor, access to the relevant systems and data, and a process owner who can define business rules and handle exceptions.
Commercially, Algebra charges a one-time implementation fee and then a fixed monthly fee to run and improve the system. Harb argues the larger investment sits on the client's side: internal agreement on who owns what, and on which measures of efficiency and revenue growth will be tracked.
The managed service is the part he treats as the product rather than an add-on.
“AI systems keep changing: models evolve, business rules shift, new exceptions appear and integrations change around them. To keep performing, a system needs ongoing care,” he says. “At Algebra, the same team designs the system, connects it to the operation, runs it in production and keeps improving it as the business grows.”
Where do-it-yourself adoption goes wrong
Harb lists four recurring errors among smaller companies:
- Buying AI as software rather than building an operating capability, with no specific business problem named.
- Automating a process before mapping its workflow, rules and exceptions.
- Ignoring how the technology and the team divide the work, when repetitive tasks should go to the system and judgement calls should stay with people.
- Treating launch as the finish line, rather than the start of monitoring, maintenance and improvement.
“My advice is to start with one clearly defined area, agree on measurable outcomes, build human oversight into the design and make someone responsible for the result. Then expand once the first system is working reliably,” he says. “The companies that benefit most from AI will not be the ones that run the most pilots. They will be the ones that turn AI into dependable operating infrastructure.”
Investors and the operator's dilemma
The four backers came together organically, Harb says, each bringing a different set of strengths. Drawing on his own experience running businesses, he names the balance between investors' long-term ambitions and the daily realities of execution as one of the harder tensions to manage. Decisions about customers, people and delivery rarely move in a straight line.
“Alignment on the destination is important, and so is the flexibility to adjust how you get there,” he says. “Clarity around decision-making matters just as much. Having experienced investors as co-founders is a tremendous advantage, and we've been deliberate from the beginning about separating strategic oversight from operating decisions.”
That separation is meant to preserve both the benefit of investor experience and the speed to build. As the company grows the arrangement will keep shifting, so the team keeps communication open and revisits expectations on a regular basis.
Beyond the Gulf
Harb's ambitions run past the region. Over the next three to five years he wants to build a global player in AI-enabled business services, with the GCC as the starting point, and to develop solutions that can be deployed across customers faster while deepening expertise in the industries Algebra already serves.
Three dos for GCC founders
- Build systems sooner than you think you need them. At Deliveroo, much of the operating knowledge sat with talented people making quick decisions. That works at small scale and becomes a brake later. Turning it into repeatable process early makes growth easier.
- Treat each Gulf market as its own. Regulation, consumer behaviour and operating dynamics differ enough that what works in Dubai may not work in Riyadh or Kuwait City. Localise.
- Adopt new technology early. Founders can design a company from day one with technology carrying work that historically required significant headcount.
Three don'ts for GCC founders
- Don't hire your way out of an operational problem. If the underlying process is inefficient, more people simply scale the inefficiency. Fix the process, then hire where people add the most value.
- Don't expand before the foundations are ready. Growth hides operational weakness, and processes that hold at small scale break quickly once new markets, products or customers arrive.
- Don't let your best people get buried in process work. The partnerships nobody pursued, the customer opportunities nobody followed up on, the analysis nobody had time to run: that is growth lost without ever appearing on a dashboard.
