Prince Mario-Max Schaumburg-Lippe: Backbase Delivery Factory Doubles Banking Delivery Speed

Most bank software projects do not slow down at the keyboard. They slow down in the handoffs. A requirement gets agreed in a meeting, written up in a document, translated into architecture, translated again into tickets, and by the time an engineer opens the file, half the original intent has quietly leaked away. Backbase, the banking operating system provider, thinks it has a fix for that, and on Tuesday, October 27, it gave the fix a name: Delivery Factory.

The idea is simple to state and harder to build. A bank delivery team works with specialized AI agents from the first agreed requirement to working code. Not one general assistant for everybody, but an agent for each job on the team. Business analysts, solution architects, delivery leads and engineers each get a partner tuned to their part of the work, and all of them, human and machine, read from the same project record.

One shared context, from meeting notes to code

That shared record is the heart of the pitch. Scoping documents, architecture specifications and meeting notes are combined with Backbase’s own product knowledge into one shared project context. Every person and every agent works from it through design, planning, implementation and review.

"A generic coding assistant starts from the code in front of it," Backbase said in its release. "Delivery Factory starts from what the bank has agreed." Code is the last step of a long chain of decisions, and an assistant that can only see the code cannot see the reasons behind it.

The agent roster is practical rather than flashy. A Business Analyst Agent drafts specifications, user stories and acceptance criteria for Jira work items. A Solution Architect Agent supports technical design decisions. Separate Backend, Mobile and Web Agents prepare implementation plans based on the project’s requirements and the bank’s own Backbase version. Engineers keep oversight of code changes and verification results, and delivery leads keep hold of scope and planning. Two further agents help banks migrate existing customer journeys, integrations and business logic onto the Backbase platform, and handle upgrades, so older work does not have to be rebuilt from scratch.

It is a pattern showing up across enterprise software right now. Readers who followed our coverage of Honeywell AI bringing a $60M upside to the Dangote refinery will spot the same logic at work: the gains come when AI is wired into the whole workflow, not bolted onto one desk.

The numbers from 12 months of real projects

Backbase is not presenting Delivery Factory as a lab experiment. The company says teams used it across more than 10 implementations over the last 12 months, and the reported results are specific enough to argue with, which is usually a good sign.

Speed, first

Teams using Delivery Factory delivered twice as fast as traditional teams. They also finished 30% quicker than teams using generic AI coding assistants. That second comparison matters more: beating a team that already has a coding assistant, by nearly a third, suggests the advantage sits in the shared context, not raw code generation.

Where the effort actually went down

The same implementations recorded 80% less effort on requirements and specifications. Read that again. Four fifths of the work of turning an agreed idea into a written, testable spec simply fell away, because the agent drafts from the shared record instead of a blank page and a calendar full of clarification meetings.

Coding and review effort dropped by 40%. Test and automation coverage rose by 60%, which may be the most telling figure of the set. Teams under time pressure usually cut testing first. A team that ships faster and tests more has not just sped up, it has changed what it can afford to do properly.

That discipline shows up in other corners of the industry, too. Our recent look at Microsoft Decision 1 scoring choices in 85 milliseconds made a similar point: speed is only worth having when the checking keeps up with it.

What Greg Fahy says the factory is really for

"Delivering banking software takes a whole team. Business analysts, solution architects, delivery leads and engineers need to stay aligned as a customer requirement becomes a working experience," said Greg Fahy, chief technology officer for digital banking at Backbase.

"Delivery Factory gives them and their AI agents a shared foundation: what the bank has agreed on, how its project is designed and how Backbase works. That helps teams spend less time reconstructing context and more time delivering the changes their customers need."

Reconstructing context. That phrase deserves to stick. Ask any delivery team where the hours go and few will say typing. They go to re-reading old decisions, chasing the current version of a document, and rebuilding in a new tool what was settled in another one three weeks ago. A system that holds the context in one place attacks the real cost.

Why banks, specifically, need this

Banking is a hard place to move fast. Backbase works with more than 120 banks across retail, small and medium business, commercial, private banking and wealth management, and most of them run core systems they cannot simply replace. Replacing a core is costly and disruptive, so banks keep systems that can be decades old and build modern experiences on top of them. That makes alignment worse, not better: more layers, more integrations, more chances for a requirement to change shape as it travels.

Delivery Factory is pitched squarely at that reality. It does not ask a bank to rip anything out. It sits in the delivery process itself, the part every bank already has, and tries to make each project arrive sooner with better test coverage than the last one.

There is a useful test readers can apply to any agentic AI launch this year, and this one passes it on paper. Does the vendor name the jobs the agents do? Yes, role by role. Does it report results from real customer work over a stated period? Yes, more than 10 implementations over 12 months. Does a human keep sign-off? Yes, engineers retain oversight of code changes and verification. And does it solve a problem teams actually complain about? Handoff drift is near the top of every delivery team’s list.

The practical takeaway

For bank technology leaders, the lesson is not to buy one more assistant for the engineering team. It is to look at where a requirement loses meaning between the meeting room and the merged code, and to give every role in that chain the same shared foundation. Teams exploring on-device and cloud combinations, like the approach in our coverage of iFlytek Spark X2.6, face the same basic question: who holds the context?

Backbase’s answer is that nobody should have to hold it in their head. Write it down once, let people and agents work from the same page, and the factory does the rest. Twice as fast, with 60% more test coverage, is a claim the next 12 months will either confirm or correct. Either way, the direction is clear, and banking teams will be watching the delivery clocks.

Also published on: