We have achieved what many thought was still 2 to 3 years away, delivering projects 2 to 5 times faster than conventional approaches, all by using AI to generate 95% of code, documentation, architecture diagrams, infrastructure as code, etc. fundamentally delivering the entire product, while increasing quality across all aspects.
Discover our methodology Gen-e2 (Generative AI Enhanced Engineering).
Across one of our offices, our engineering team sits huddled around a large screen for an 'AI mob programming' session. There's no frantic typing, no endless scrolling through libraries of code on Stack Overflow. Instead, we're engaged in an animated discussion about the business logic of a project, while our engineers prompt GitHub Copilot to generate the entire architecture and code of the product in real-time.
This is how we now build software at PALO IT, and it's revolutionizing our entire approach to software engineering. But getting here wasn't straightforward, and our journey taught us valuable lessons about the future of software engineering.
From a problem, to a breakthrough
It started with a problem. Like many organizations, we initially deployed AI coding assistants to augment our standard Agile Software Development Life Cycle (SDLC), expecting significant productivity gains. The results were underwhelming because we were trying to fit AI into an outdated delivery paradigm. We realized that using AI tools at the individual level, without rethinking the orchestration of the entire chain of production, was limiting productivity gains to the individual level, and only impacting development time within silos. The breakthrough came when we understood that meaningful transformation required reimagining the entire software development lifecycle to be 'AI-first' or 'AI-driven', instead of AI just augmenting standard delivery methodologies.
Through 18 months of experimentation and enterprise projects, we have developed and validated the right approach, entire libraries of prompts, configuration files, and the repository structure to fundamentally change how AI generates a product. At its heart is a way to make AI maintain a comprehensive understanding of the entire project context. Every decision—from business goals the product is meant to achieve, to high-level architecture choices, to specific implementation details—is documented, validated, and traceable. Gen-e2 doesn't just generate code, it generates systemic understanding of the product.
We use the context to prompt, and we prompt to create context
Instead of prompting to generate small features, we prompt the AI to generate a wider understanding of what the product needs to achieve, so that in turns the AI generates avenues and pathways to build this product. We do that by going beyond “Chain-of-Thought” prompting, which is what 80% of people do – basically a linear way of jumping sequentially from questions to answers, prompt after prompt.
Fig 1. “Chain-of-Thought” prompting: linear sequential prompts in a succession of question-answer-question-answer
With Gen-e2, we use “Context Prompting” instead. The context of the entire project expands from the prompts. In turn, because the AI holds this entire context, we can use much smaller prompts: i.e. engineers don’t have to constantly repeat themselves and write lengthy prompts. The AI understands what they mean through shorter prompts, creating significant productivity gains. The project context becomes a complex adaptive system made of feedback loops that keep enriching it and generating systemic understanding.
Fig 2. Gen-e2 Context Prompting concept
How Gen-e2 delivers a new product development lifecycle
What makes our Gen-e2 approach unique is that we're not just theorizing about the future of AI-driven development — we're implementing it at scale in ways that even the most mature AI teams say they haven't seen before. We typically start with a pilot project to prove our Gen-e2 AI-PDLC (Product Development Life Cycle) works within the organization’s context. Even those who have been pushing AI copilot tools as early adopters say they want to partner with us when they realize that we are executing on AI adoption levels they had only imagined hypothetical, or years away for their own organization.
This response from key organizations, forging ahead with AI, confirms what we've discovered through practical experience: Gen-e2 represents a fundamental leap forward in how AI can be integrated into the software engineering process, and probably the most significant change in this space over the last decade.
The impact on our project team roles has been transformative. It is not just engineers who work with AI copilot tools, but also product owners, designers and architects. Rather than making developers obsolete, Gen-e2 has elevated their role. Our engineers now spend more time on valuable, strategic work, focusing on business outcomes rather than implementation details. Through our AI-enhanced mob programming sessions, our product teams work in unprecedented alignment with business stakeholders, conducting rapid one-day iterations that bring together business strategy and technical execution.
The numbers that change everything
The results speak for themselves! We're delivering new products 2-3 times faster than traditional methods, while legacy systems modernization and re-platforming is seeing a 3-5 times acceleration.
Fig 3: Traditional Software Development Life Cycle, despite the agile intent, still disparate, fragmented phases, leading to much longer delivery cycles
In practice, this is what the burn up chart of a project we recently delivered using GitHub Copilot at a major international organization looks like, on Fig. 4. The velocity has been multiplied by two thanks to Gen-e2 (compared to traditional methods) and we have been able to deliver additional features to reach 120% of the initial scope.
Fig 4: Burn up chart of a project recently delivered at a major international organization, Gen-e2 actuals vs. forecast following traditional methods, increased velocity and alignment through rapid one-day iterations in mob-programming sessions between product owners, designers and engineers
As AI technologies progress globally, by the week, we are already pushing beyond the basic use of AI copilots, and are now coupling them with specialized 'AI agents' to perform architecture, design, and compliance tasks. Gen-e2 isn't just changing how we work at PALO IT — it's making software development more accessible to organizations of all kinds, shifting focus from writing code to solving business problems.
As one of our engineers observed, "Code is from the past. We're not just writing code anymore. We're engineering solutions." Traditional phases haven’t disappeared simply because AI is faster — they have vanished because our entire development process has become more intelligent, integrated, and aligned with business needs. In this new paradigm, every line of code is generated with purpose, every architectural decision is made with full context, and every team member focuses on what truly matters: creating value for users. We believe this represents the future of software and product engineering — not AI replacing humans, but AI helping humans work at a higher level of abstraction, thinking strategically while machines handle the tactical details. Based on our experience, this future isn't just possible — we're already living it.
If you're ready to transform your software development process and unlock unprecedented productivity, contact us today.
– Dimitri Baikrich, CTO, PALO IT