The COBOL Colleague Difference

Evolve your mainframe application portfolio without risking your business

COBOL Colleague powers startling productivity on the mainframe.

COBOL Colleague evaluates source code like a developer but with perfect and complete understanding of all the logic, all at once.

But it’s not just about productivity. The developers that grok your age-old, critical applications, the coding heroes with all the stories, the saviors with million-dollar solutions … these people are going, going, gone. New developers don’t have time to spend years learning the application.

You need a strategy that allows your mainframe application portfolio to evolve without risking your business. A different kind of platform, built with a different kind of AI, Colleague is based on years of research in cognitive automation that culminates in a collaborative experience different from every other tool designed for developer productivity.

You care about the business, not just the code

It's easy to get lost in the constructs of code and language. Your mainframe portfolio is there for one purpose – to run your business. So why do existing tools focus on the code and not the business intent of the application?

Current AI development productivity tools rely on pattern matching across large repositories. COBOL Colleague understands code the way your developers understand code – as behaviors that serve the business.

Imagine a musician in contact with an instrument never-before-seen. He keenly experiments with its keys, strings, valves, slide, and pedals – its whatchamacallits – to determine the sound production. Is it a rhythm, reed, wind, or string instrument, or even one of an unknown type?


Then watch a musician accomplished with this instrument. There is a simplicity and a visceral response to her handling of the mechanics. She demonstrates the moving pieces with expertise, producing music. The observer’s mind goes from heightened inquisitiveness to one of catharsis, knowing and finding satisfaction in her knowing – a sudden familiarity with a spark of delight.

Instead of returning a set of sporadic text matches, COBOL Colleague returns the code that encompasses the business process. Just what the developer was looking for, and FAST!


Find the business functionality, not just words

There are sophisticated tools to search across enterprise programs with millions of lines of code. What developers find with them are matching words and phrases. They paste them in other tools and search for them further.

Words in applications are road signs, often cryptic and fiercely limiting in communicating logic. When variables and tables and programs change names entirely, without changing meaning, text search can be confusing, even overwhelming.

What about business rules and processes, and software requirements? These are the things that make systems valuable and make your business run.

The crucial concepts, the enterprise business knowledge that is hard to unravel, are buried between the words. Returning strings in characters is like returning page numbers in a book when a reader just wants to know the plot and the storyline.

Utilizing intelligence augmentation, COBOL Colleague extrapolates the business functionality in enterprise applications, not just words, and then makes complex, encoded logic, obvious. This is significantly more than search. It is an interactive application show-and-tell, that operates at scale with perfect knowledge.

Ease the cognitive load through intelligence augmentation

Developer tools automate repetitive tasks – like searching and diagramming relationships, manually stepping through logic, pointing out patterns that have been errant in other applications, and so on. They all require experience, and often a third language, to interpret the output. Simpler, but still with an inordinate cognitive load. Hardly 21st-century technology.

The real limitation to understanding the complexity of your applications is the amount of logic that a developer can simulate in their cortex – their cognitive load. The Colleague platform tackles that limitation head-on with intelligence augmentation.

It has already analyzed, observed, and found the relationship of every piece of logic – before the developer has even opened the tool.


The developer provides the criteria, and Colleague returns the tome of answers. With a few clicks, 1MM lines of code is reduced to just the relevant business behavior – isolated from the noise so the developer can easily understand and implement their change. Now that’s true productivity.


Active Agent is an expert team player

COBOL Colleague showcases a new-age user experience including Active Agent -- a guide designed to know and navigate your entire portfolio, under the control of your developers.

The developer provides the criteria, step by step and in response, the Active Agent seamlessly guides the user, distilling the code into a single, perfectly organized, comprehensible block of code.

With Active Agent, your developer doesn’t need years of expertise. They don’t need an enterprise expert in your shop to show them the applications, and they don’t need training to understand the results of a tool.

Active Agent is the expert team player, making suggestions based on context and the line of questioning – guiding the user as a world-class coach with impeccable wisdom and a perfect track record. No cut and paste just a collaborative conversation, each partner focusing on what they do best. The outcome is the code the developer needs to perform their task, and nothing else.

Precise from every angle

COBOL Colleague knows the dependency of every line of code on every other line of code. The result – it knows the boundaries of every possible value and all the logic in between.

Other tools require massive data sets, production data, and hypotheses to identify errant behavior that is bound to data. Colleague already knows every possible outcome of any data input with no file or database at all, no runtime, just source code.

The answer is embedded in your code and can be found easily, without running your application or sampling hit/miss data sets where testing had failed. This means expeditiously isolating behaviors, predictably, with way less thought.

Active Agent knows the range of values of every variable as they flow through every table, every file, every computation, and assignment … literally everything about the data from input to output.


"The ratio of time spent reading versus writing [code] is well over 10 to 1."

Robert C. Martin
Author and Agile Advocate

Propel the maintenance, evolution, and transformation of your programs with COBOL Colleague

Get a peek into the future of AI-assisted software development!