Posts Tagged ‘ContextEngineering’
[VoxxedDaysAmsterdam2026] Coding Fast and Slow: Managing Mental Energy for Sustainable Software Development
Lecturer
Baruch Sadogursky is a Developer Advocate at Tessl, focusing on package management for AI agent skills. A Java Champion with extensive experience in DevOps and software engineering, Baruch explores the intersection of behavioral psychology and programming practices to improve developer effectiveness and well-being.
Abstract
Software development demands significant cognitive resources, yet the mental costs of context switching, interruptions, and decision fatigue are frequently overlooked. Baruch Sadogursky applies insights from behavioral psychology, particularly Daniel Kahneman’s framework of fast and slow thinking, to examine how developers can manage mental energy more effectively. The presentation analyzes the biological and psychological mechanisms underlying attention, fatigue, and intuition, while offering practical strategies for preserving cognitive capacity throughout the workday. Topics include reducing unnecessary context switches, implementing deliberate work patterns, and leveraging AI tools with appropriate context engineering. These approaches enable developers to maintain high-quality output while avoiding burnout and sustaining long-term productivity.
The Dual Systems of Human Cognition in Programming
Human decision-making operates through two distinct cognitive modes. System one functions automatically, emotionally, and with minimal effort. It excels at pattern recognition and rapid responses but can lead to errors when complex analysis is required. System two engages deliberate, logical, and effortful thinking. It handles complex problem-solving and critical evaluation but consumes significant mental energy and operates more slowly.
In software development, system one drives much of routine coding activity. Experienced developers intuitively navigate familiar codebases, recognize common patterns, and make quick implementation decisions. This efficiency feels productive and satisfying. However, system one also introduces risks. Code that “looks okay” may contain subtle flaws that system two would identify through careful review. When mental resources are depleted, developers default to system one even for tasks requiring deeper analysis, resulting in overlooked issues and technical debt.
System two becomes essential for architectural decisions, debugging complex interactions, code reviews, and learning new technologies. The challenge lies in preserving sufficient system two capacity for these critical activities rather than exhausting it on routine interruptions and low-value tasks.
The Hidden Costs of Context Switching and Interruptions
Modern work environments are engineered to fragment attention. Email notifications, messaging platforms, meetings, and status updates create frequent context switches throughout the day. Each interruption forces the brain to reload relevant information, reestablish mental models, and regain focus. This process is metabolically expensive and significantly reduces overall effectiveness.
Research demonstrates that developers check email approximately 77 times daily on average. Attention spans on digital interfaces average just 47 seconds before shifting to another task. Returning to deep work after an interruption requires substantially more time than most people realize. The cumulative effect is reduced code quality, increased errors, and diminished creative problem-solving capacity.
Critically, developers often fail to recognize when their cognitive performance has declined. Similar to how tired individuals overestimate their driving ability, programmers working on system one may believe their code quality remains high. This self-assessment bias makes proactive management of mental energy essential rather than reactive.
Strategies for Preserving and Replenishing Cognitive Resources
Effective mental energy management requires both prevention of depletion and strategic restoration. Time blocking dedicates specific periods to focused work without interruptions. The Pomodoro technique, originally using 25-minute intervals, establishes minimum focused periods while allowing flexibility for natural flow states. When developers enter deep concentration, they should continue beyond the timer rather than forcing artificial breaks.
Task batching groups similar activities to minimize context switches. Responding to all messages during designated periods rather than reacting immediately preserves cognitive continuity for primary development work. Physical exercise, adequate sleep, and mindfulness practices support overall cognitive resilience. Sleep in particular serves as the primary mechanism for restoring system two capacity, making consistent rest non-negotiable for sustained performance.
Workspace organization and notification management reduce environmental triggers for attention shifts. Tools that intelligently manage calendars and protect focus time help implement these practices at scale. Delegation of routine tasks to appropriate automation or AI assistance frees cognitive resources for higher-value activities.
Engineering Context for Human and Machine Systems
Context engineering addresses both human and artificial intelligence systems. For developers, clear documentation, consistent coding standards, and well-structured codebases reduce the mental effort required to understand and modify systems. When context is preserved, system two can focus on creative problem-solving rather than basic comprehension.
Artificial intelligence systems similarly require rich context to perform effectively. Large language models benefit from detailed prompts, relevant examples, and domain-specific knowledge. The same principles that help human developers maintain context—clear boundaries, explicit documentation, and systematic organization—enhance AI performance when integrated thoughtfully.
Modern AI coding assistants represent both opportunity and responsibility. When used with appropriate context, they augment human capabilities without replacing critical thinking. However, over-reliance on AI without sufficient system two engagement can lead to acceptance of suboptimal code that appears functional but lacks deeper quality. Balancing AI assistance with human judgment remains essential for maintaining code integrity.
Building Sustainable Development Practices
Organizations can support cognitive sustainability through intentional practices. Protected focus time, reduced meeting loads during deep work periods, and recognition of the costs of context switching contribute to better outcomes. Engineering workflows that minimize unnecessary interruptions while maintaining necessary collaboration create environments where developers can consistently access their full cognitive capabilities.
Individual developers benefit from self-awareness of their energy patterns and implementation of personal systems for protection and restoration. Regular reflection on work patterns, experimentation with different techniques, and honest assessment of code quality under varying conditions build more effective personal practices.
The goal extends beyond short-term productivity to sustainable careers in software development. By treating mental energy as a finite and valuable resource, developers and organizations can achieve higher quality outcomes while reducing burnout and turnover.
Technology continues to evolve, with AI assuming more routine tasks and shifting human roles toward system-level thinking and creative problem framing. The ability to manage cognitive resources effectively becomes increasingly valuable in this landscape. Developers who master these skills will maintain their effectiveness and competitive advantage regardless of technological changes.
Conclusion: Toward More Conscious Software Creation
Software development is fundamentally a cognitive activity requiring sustained mental effort and clear thinking. Understanding the mechanisms of attention, fatigue, and decision-making empowers developers to work more effectively. By applying principles from behavioral psychology to daily practices, teams can reduce unnecessary cognitive costs and preserve capacity for the most valuable aspects of their work.
The combination of thoughtful process design, appropriate tool usage, and individual self-management creates conditions where developers can consistently produce high-quality work while maintaining their well-being. In an industry characterized by rapid change and high demands, these skills become essential for long-term success and satisfaction.
Conscious management of mental energy transforms software development from a reactive, exhausting process into a sustainable, engaging craft. The investment in understanding and optimizing cognitive performance yields returns in code quality, innovation, and professional fulfillment.
Links:
[AWSReInvent2025] The Agentic Frontier: Lessons from Anthropic’s 2025 AI Deployments
Lecturer
Danny Leybovich is a Product Lead at Anthropic, dedicated to building the infrastructure and models that empower the next generation of AI developers. With a focus on high-reasoning models and developer experience, Danny has been instrumental in the launch of Claude Code and the evolution of Anthropic’s agentic framework. His work centers on the practical realities of moving AI from “cool demo” to “reliable autonomous system.”
Abstract
2025 marked a pivotal shift in the artificial intelligence landscape: the transition from interactive chatbots to autonomous AI agents. This article synthesizes the key discoveries made by Anthropic during this transformative year, particularly through the development of Claude Code and the deployment of the Opus 4.5 frontier model. It explores the “agentic architecture” required for long-horizon autonomous work, emphasizing the critical roles of context engineering and skill acquisition. The analysis examines the shift toward “agent-first” workflows, where the model is no longer a passive assistant but an active participant with multi-hour reasoning capabilities. By investigating patterns of reliability and the evolution of AI engineering practices, this article provides a roadmap for the next wave of agentic AI.
The Shift to Agent-First Workflows
In the early stages of generative AI, the predominant interaction pattern was the “chat” interface—a stateless exchange where a human provided a prompt and the model provided a response. 2025 saw the obsolescence of this limited model in favor of “agent-first” workflows. In an agentic architecture, the model is granted the autonomy to use tools, manage its own memory, and pursue goals over extended periods—sometimes lasting hours.
This shift changes the fundamental role of the developer. Instead of engineering a single prompt, the developer now engineers an environment in which an agent can succeed. This involves defining clear objectives, providing access to necessary APIs, and implementing “guardrails” that ensure the agent remains on track during autonomous loops. The rise of “Claude Code”—an agent that can autonomously file GitHub issues and build applications—serves as the flagship example of this transition.
Advanced Context Engineering: Beyond the Context Window
While early AI discussions focused heavily on the size of the “context window,” Anthropic’s experience in 2025 highlighted that quality of context is far more important than raw volume. Context engineering is the practice of strategically selecting and formatting the information provided to the model to maximize reasoning accuracy and minimize hallucinations.
Effective context engineering for agents involves:
- State Management: Keeping track of what the agent has already done and what remains to be accomplished.
- Relevant Document Retrieval: Using RAG (Retrieval-Augmented Generation) to pull only the most pertinent information into the reasoning loop.
- Semantic Chunking: Ensuring that the information is presented in a way that the model can easily digest and connect to other data points.
By focusing on context engineering, developers can enable agents to maintain “state” across long horizons, allowing for complex tasks like refactoring an entire codebase or conducting multi-step regulatory research without losing the thread of the original objective.
Tool Construction and Skill Acquisition
A primary differentiator for AI agents is their ability to interact with the world through tools. In 2025, Anthropic refined the methodology for “teaching” agents new skills through tool construction. A “skill” is essentially a well-defined tool—such as a Python interpreter, a SQL query engine, or a web search function—that the model knows how and when to invoke.
The engineering challenge lies in creating “reliable” tools. If a tool’s output is ambiguous or inconsistent, the agent’s reasoning loop will break. Therefore, tool writing has become a core discipline within AI engineering. Developers must create tools that provide “structured feedback” to the model, allowing the agent to self-correct if a tool call fails. This iterative loop of tool use and self-correction is what allows agents to handle “long-horizon” tasks that were previously impossible for LLMs.
Analyzing the Performance of Opus 4.5
The release of the Opus 4.5 frontier model provided the reasoning “horsepower” necessary for the agentic revolution. Unlike smaller models that might prioritize speed, Opus 4.5 is optimized for high-reasoning tasks. Its performance characteristics include a significant reduction in “logic drift”—the tendency of a model to lose focus during long sequences of thought.
In production environments, Opus 4.5 has demonstrated an ability to navigate “deep” decision trees. For example, when tasked with finding a bug in a complex software system, the model can formulate a hypothesis, write a test to prove it, analyze the test results, and then iteratively refine its approach. This capability for “autonomous debugging” is a hallmark of the newest wave of AI, where the model’s intelligence is leveraged not just for text generation, but for problem-solving in dynamic environments.
Code Sample: Defining a Secure Tool for Claude Agentic Workflows
'''
Conceptual tool definition for an Anthropic Agent
This tool allows the agent to safely query a database
'''
def get_tool_definition():
return {
"name": "query_database",
"description": "Allows the agent to execute read-only SQL queries to retrieve customer data.",
"input_schema": {
"type": "object",
"properties": {
"query": {
"type": "string",
"description": "The SQL query to execute. Must be read-only."
},
"max_rows": {
"type": "integer",
"default": 10
}
},
"required": ["query"]
}
}
'''
This structure enables the model to 'reason' about when it needs
to fetch data versus when it can rely on its internal knowledge.
'''
Long-Horizon Autonomous Reliability
The final frontier explored in 2025 was the challenge of reliability. For an agent to be truly useful, it must be able to work for hours without human intervention. This requires a robust infrastructure that can handle model timeouts, API failures, and unexpected edge cases.
Anthropic’s research into long-horizon agents suggests that reliability is not a feature of the model alone, but a result of the model-infrastructure synergy. This includes:
- Checkpointing: Periodically saving the agent’s state so it can resume after a failure.
- Human-in-the-Loop (HITL) Triggers: Designing the agent to “ask for help” when it reaches a confidence threshold that is too low.
- Verification Loops: Implementing a secondary model or a deterministic process to verify the agent’s output before it is committed.
These patterns are what define the current state of the art in AI engineering, moving the industry toward a future where agents are trusted partners in the enterprise.
Conclusion
The lessons of 2025 are clear: the future of AI belongs to autonomous agents. By mastering the disciplines of context engineering, tool construction, and long-horizon reliability, developers can leverage models like Claude Opus 4.5 to solve problems of unprecedented complexity. As we look ahead, the trends established this year—particularly the move toward agent-first workflows—will define the next decade of technological innovation. The demo era is over; the production era of agentic AI has begun.