LSA mindset in AI era
For decades, the Lead System Architect (LSA) has guarded technical excellence by designing systems that scale, adapting them as business needs change, and earning stakeholder trust. These responsibilities remain, but the role continues to evolve.
Generative and agentic AI have blurred the once-clear line between business logic and technical implementation. AI no longer just runs tasks at the edges of a process. It works inside the Case, reasoning and collaborating with people. Business users describe what they need in plain language, and AI turns it into a starting design in minutes.
The real question is: When AI speeds up everything, where do you add value?
You own the business outcome, not just the solution. Solving the stated problem is not enough. Your solution must deliver measurable business value and an effective experience for the people who use it.
AI now helps produce both designs and working solutions, so the gap between design and build has narrowed. Applications also connect in more ways than before: through standard APIs and now through agent protocols. The Model Context Protocol (MCP) connects AI agents to tools and data, while the Agent2Agent (A2A) protocol supports collaboration between agents.
Your job is to understand the business context, choose the solution that best serves the outcome, and take responsibility for it.
What it takes to design for business outcomes
As the LSA on a Pega implementation, you shape every factor that affects the business outcome. Focus on the following areas:
Choose the best solution and justify it. AI can propose many solutions to the same problem, and even well-written prompts can return different answers each time. Compare the options based on performance, scalability, security, user experience, reusability, and maintenance effort. Write down why you recommend one, so the business can see its value and approve it with confidence.
Design for the whole application lifecycle. Your job does not end when development ends. Design an application that stays effective throughout its life in performance, scalability, and maintenance. Follow the Build for Change® principle: keep the design flexible enough to handle change, but do not build for guessed future needs. Speculative features weaken what the business needs today.
Optimize time, cost, and effort. Technical teams often overlook these factors, but the business never does. Weigh them across design, build, and daily use. AI can speed up outcomes, but it brings token costs, effort to refine its output, and time to upskill the workforce.
Pega Infinity™ 26 tackles token cost through the Pega Predictable AI™ architecture. It uses AI reasoning once at design time: Pega Blueprint AI™ and Pega Infinity Studio™ help your team design the best agentic workflows. At run time, agents do not reason through the process again for every request. A lightweight AI query works out what the user wants and picks the right Pega workflow, and the agent then follows that workflow step by step. When a step needs deeper large language model (LLM) work, such as parsing a document, you give it specific, limited instructions.
The result is predictable outcomes and predictable cost. As an LSA, use this design: put AI reasoning into well-designed workflows, and use LLMs at run time only for steps that require them.
Additional best practices include:
- Make user experience a top priority. Teams often treat user experience as an afterthought in business-to-business and internal applications. But business users drive productivity, and productivity depends on seamless navigation and how easily users finish their tasks. Move user experience from the bottom of your list to the top.
- Decide where AI belongs, and where it does not. Not every step needs an agent. Use AI where judgment, language, or reasoning adds value. Use rules and workflows where the business needs predictable, auditable results. Decide where a human must review or approve.
- Build trust into the design. Plan for guardrails, explainability, auditability, and data privacy from the start, not after go-live. Without them, the business will not trust AI-driven work.
- Define and measure the outcome. Agree on success metrics and key performance indicators (KPIs) with the business before you design, and track them after go-live. Then you can show the value instead of just claiming it.
- Lead your team through the change. Coach your developers to use AI tools well and to treat AI output as a draft: review it, test it, and take ownership of it.
The LSA is the orchestrator of business outcomes
AI can accelerate design, assist with logic generation, and automate workflow execution. But AI does not own a business outcome. It does not know which trade-off a client can afford, which shortcut becomes tomorrow's technical debt, or which design will continue to perform as volumes increase. That judgment comes only from experience, and that experience belongs to the LSA.
You have spent years designing enterprise applications, solving complex problem, and earning stakeholder trust. That background puts you in a unique position. You can turn AI's speed into business value, choose the right AI solution for the business need, and hold your team accountable for what reaches production.
Your role has not shrunk in the AI era. It has grown. You are no longer just the guardian of technical excellence. You are the architect of business outcomes. Design every solution for the outcome that it must deliver, prove its business value, and own that outcome from the first blueprint to the last enhancement.
Check your knowledge with the following interaction:
This Topic is available in the following Module:
Want to help us improve this content?