Core Philosophy
Critical Product Judgment
instead of: Faster AI Slop
As AI collapses the distance between product, design, and engineering, companies will need fewer translators and more people who can own the whole system. Product Architects that understand what should be built, why it matters, how it gets built, and what happens if it fails.
Software is getting cheaper to produce. That abundance is a trap if teams confuse speed with leverage. I support an AI-native mindset, but will always bet more for human amplification over full automation.
Because in the Industrial AI age, the scarce resource will always be critical judgment: knowing which problems are worth solving, especially in environments where decisions can carry real-world consequences.
- Daniel Arevalo, June 2026
How I Work
Product architecture in the Industrial AI age
My skills enable teams to execute fast, correct direction, align quickly, and iterate with high velocity towards practical solutions and the right product bets.
Strategy
What is worth building
As a product architect, I turn research, stakeholder tension, and domain constraints into a clear call on which features deserve to exist when coding is no longer the bottleneck.
Exploration
How it gets built
I prototype in code, sit across design and engineering, and collapse the distance between decision and execution. Judgment without the ability to build is only half the job.
Execution
What happens if it fails
Someone still has to care about the outcome. I keep monitoring product features through launch and use, validating whether each works, and learning when it is wrong and why.
Speaking & Advocacy
Opinions on product, taste, and judgment
- Event RunwayFBU Takeoff The Future of Product in the Industrial AI Age
- Event Tekna AI & Cybersecurity Operationalizing AI for Engineers
- Event RunwayFBU Founder Events The Designer's Role in High-Stakes Tech
- Event ProductTank Riga Amplification vs Automation Architecture
Technical Stack
AI-Native Tools and Agentic Workflows
My workflow pairs product judgment, fast iterations and production-ready prototyping. I use AI coding agents to ship faster while keeping ownership of the decisions that are made with any combination of people and AI. I like to share what I learn from these loops as the tools change.
All the agentic loops and graph engineering are there to guarantee one single thing:
The better the context that goes in, the better result that comes out.
Autonomous research agent for multi-step investigation, browsing, and turning open questions into structured output.
Cited search, literature review, and fast knowledge extraction with source-grounded answers.
Agent and MCP-oriented workflows for structured experiments, tool integration, and reproducible research loops.
IDE-native coding assistance, completions, and refactors grounded in the codebase and workflows.
Source-grounded notebooks with long-context reasoning for structured synthesis and stakeholder alignment.
JTBD and opportunity framing turned into stakeholder-ready decks, narratives, and decision artifacts.
Code generation, architecture support, and technical documentation.
Cloud workspaces and instant deploys for quick experiments, MVPs, and shareable demos.
Agent-built full-stack apps for fast iteration on flows, UI, and interactive product experiments.
Prompt-to-full-stack builds for polished UI prototypes and rapid user validation.
Agent-first editor for repo-wide context, inline AI, and fast iteration on production code.
Agentic terminal for repo-aware commands, multi-step agent runs, and reviewable changes without leaving the shell.
Design context, components, and handoff in the agent loop via Model Context Protocol.
Issues, projects, and roadmaps available to agents for planning and delivery workflows.
Local Postgres, migrations, and project scaffolding from the terminal for fast backend iteration.
Webhook forwarding, test charges, and billing flows in local and staging environments.
The tools speed up the previously busy work in the middle of the product development process. The leverage for talented PMs and designers still comes from knowing which loop is worth running — and stopping when the output is slop.
Read more about my work
Design, AI & Product Thinking
- Intent engineering for AI agents
- About craft, taste, and speed
- The disappearing middle of software work
- Design leadership has evolved
- The death of front-end development
- How I write code in 2026