Practical essays on AI, product, and SMB growth.
Plain-spoken, opinionated, and written for the people doing the work.
REA Tools and the Five-Step Pipeline for Reverse-Engineering Applications (Part 2 of 5)
Part 2 of 5. The REA tool chain and the five-step pipeline — deconstruct, analyze, visualize, insight and roadmap, enhance — plus the six benefits that make REA a weekly discipline, not a specialist skill.
AI StrategyWhat Is Reverse Engineering, and What Is REA? (Part 1 of 5)
Part 1 of 5. Reverse engineering recovers architecture from a binary without source. REA — Reverse-Engineer Anything — extends that same discipline to legacy systems, dependencies, competitors, incidents, and markets.
AI StrategyIs There a Clear Winner Among Opencode, Claude Code, Codex, and Cursor? (Part 15 of 15)
Part 15 of 15. The verdict after fourteen parts of comparison: no overall winner, but a clear per-scenario winner — Cursor for solo builders, Claude Code for mid-market teams, Opencode for self-host, Codex for background runs.
AI StrategyUnique Strengths of Opencode, Claude Code, Codex, and Cursor (Part 14 of 15)
Part 14 of 15. One sentence per tool: the real differentiator that remains after thirteen parts of comparison — model freedom for Opencode, long-context plans for Claude Code, cloud background runs for Codex, in-editor Tab for Cursor.
AI StrategyCode Accuracy of AI Coding Tools: Real-World Performance (Part 13 of 15)
Part 13 of 15. The six axes of real-world code accuracy for Opencode, Claude Code, Codex, and Cursor — compile rate, test-pass rate, hallucination rate, semantic correctness, security defects, and the benchmark-vs-reality gap.
AI StrategyCollaboration Features in AI Coding Tools: Teamwork Compared (Part 12 of 15)
Part 12 of 15. How Opencode, Claude Code, Codex, and Cursor actually support team coding — PR integration, shared prompts and rules, admin and billing, review loops, and what they do not do.
AI StrategyProgramming Languages Supported by AI Coding Tools (Part 11 of 15)
Part 11 of 15. Which languages Opencode, Claude Code, Codex, and Cursor actually handle well — the tier list, the long tail, and why the real question isn't the language, it's the tooling around it.
AI StrategyAPI Integration and Workflow Compatibility: AI Coding Tools Compared (Part 10 of 15)
Part 10 of 15. How Opencode, Claude Code, Codex, and Cursor plug into your existing stack — APIs, IDE integrations, CI, data residency, and the switching cost each one carries.
AI StrategyCode Generation Speed and Efficiency: Opencode, Claude Code, Codex, Cursor (Part 9 of 15)
Part 9 of 15. The six real speed-and-efficiency metrics for AI coding tools — first-token latency, task completion time, token efficiency, cost per task, review burden, and code quality — scored across Opencode, Claude Code, Codex, and Cursor.
AI StrategyNLP and Context Understanding in AI Coding Tools (Part 8 of 15)
Part 8 of 15. The six NLP capabilities that decide whether an AI coding tool actually understands your prompt — and how Opencode, Claude Code, Codex, and Cursor stack up on each.
AI StrategyCode Synthesis Techniques: What Each AI Coding Tool Actually Does (Part 7 of 15)
Part 7 of 15. The six synthesis techniques that separate Opencode, Claude Code, Codex, and Cursor — context assembly, planning, tool use, self-correction, model routing, and diff format.
AI StrategyCode Completion Compared: Opencode, Claude Code, Codex, Cursor (Part 6 of 15)
Part 6 of 15. How Opencode, Claude Code, Codex, and Cursor stack up on four code-completion modes — inline tab, chat edits, agentic multi-file edits, and autonomous background tasks.
AI StrategyTraining Data: How It Shapes AI Code Accuracy and Quality (Part 5 of 15)
Part 5 of 15. Where Opencode, Claude Code, Codex, and Cursor get their training data, how quality beats quantity, and the four signals you can actually verify before trusting a model with production code.
AI StrategyDeep Learning Foundations Behind AI Code Generators (Part 4 of 15)
Part 4 of 15. The transformer architectures, pretraining corpora, fine-tuning passes, and reinforcement-learning loops that make Opencode, Claude Code, Codex, and Cursor possible — in plain English.
AI StrategyThe Role of Language Models in AI Coding Tools (Part 3 of 15)
Part 3 of 15. How large language models actually power Opencode, Claude Code, Codex, and Cursor — what the model does, where the tool starts, and why that line matters for every pick in the series.
AI StrategyClaude Code vs Codex vs Opencode vs Cursor: An Overview (Part 2 of 15)
Part 2 of 15. A plain-English overview of the four leading AI coding tools in late 2026 — Claude Code (Anthropic), Codex (OpenAI), Opencode (open-source), and Cursor (AI-first IDE) — with a side-by-side comparison table.
AI StrategyAI Code Generation and Developer Productivity: Introduction (Part 1 of 15)
Part 1 of a 15-part series comparing Opencode, Claude Code, Codex, and Cursor. What AI code generation actually is in late 2026, where the productivity gains come from, and how to read the comparison that follows.
AI StrategyChoosing a Decision API in 2026: My Practical Framework
Thirteen decision APIs, five questions, five short picks. The five-question framework a Fractional CTO uses to pick a decision API vendor in late 2026 — plus why Jev stays the default while I A/B test the alternatives.
AI StrategyWhere Decision Models Fit in an AI Agent Stack
A fast decision model belongs in front of expensive LLM calls — routing, tool selection, loop-break and escalation. The two-speed agent pattern, with a reference architecture builders can copy.
AI StrategyOpen-Weight Decision Models: Can You Run Your Own Jev?
Kev, Laya, Tev1 and Cloudflare's Clef put open-weight decision models in reach. Here's the honest tradeoff — privacy and control vs GPU cost, ops work, and license fine print — and the volume math that decides whether self-hosting pays.
AI StrategyCalibration Is the Feature Nobody Is Talking About
The decision model industry has started a benchmark war on accuracy. Meanwhile the number that actually determines whether you can automate — calibration — barely gets a slide. Here is why calibration, not raw accuracy, is the under-discussed quality bar in 2026.
AI StrategyBe Careful With 'Beats Jev': How to Read Decision Model Benchmarks
Most 'beats Jev' claims come from the vendors themselves, often on benchmarks the model was trained near. Here's a 2026 field guide to decoding decision model benchmark charts — plus a one-afternoon protocol to test any vendor on your own data.
AI StrategyThe One-Line Migration: Which Jev Competitors Accept the Same Requests
Liquid AI's d1, Nace.AI's Drex, Celeris-1, and Upstage's Solar Decide all accept TypeSafe Jev's request format. Why that wire compatibility makes A/B testing vendors almost free — and how to build a provider-abstraction layer so you are never locked in.
AI Strategy13 Decision APIs in 3 Weeks: A Builder's Map of the Market
From TypeSafe's Jev to Cloudflare's Clef, OpenAI, Microsoft, Liquid AI, Upstage, AWS, and smaller open-weight entrants — a market map of the decision API category and why it commoditized the moment it was created.
AI StrategyThree Weeks From Launch to Clone: OpenAI Enters the Decisions Market
OpenAI previewed its Decisions API at DevDay 2026, fourteen days after TypeSafe AI launched Jev. A deep-dive on GPT-6 Luna's pricing, image input, and what the clone timing says about the decision model category.
AI StrategyWhat Is a 'System One' Model, and Why Should Builders Care?
A System One model is an AI tuned to decide, not deliberate — a label, score, or probability in milliseconds. A plain-English explainer for builders, with the Kahneman origin and when it beats an LLM calling structured outputs.
AI StrategyI Stopped Asking My AI to Write. I Started Asking It to Decide.
A first-person account of the mindset flip that reshaped my book platform: moving five structured judgments from a chat LLM to a decision model, and what changed in how I design AI calls.
AI StrategyGetting Started: Your First Decision Model Experiment
A coach-walks-you-through-it guide to your first decision model experiment in a single day: pick a repetitive task, write the question, and test on 20 real examples before you invest further.
AI StrategyDo You Need to Be a Developer to Benefit From This?
A plain-English guide for non-technical founders, ops leaders, and product managers: how decision APIs actually work, how no-code tools expose them, and the questions to ask your tech team or vendor before you buy.
AI StrategyOpenAI, Microsoft and Cloudflare Joined In: Why That Matters
Within 24 days of TypeSafe AI's Jev launch, OpenAI, Microsoft, and reportedly Cloudflare all shipped their own decision APIs. Here's why that speed signals category validation — and why builders should treat decision models as infrastructure, not an experiment.
AI StrategyHow I Use a Decision Model in My Own Product
A field report from my book platform: three places I replaced chatty AI calls with Jev's typed decisions — quote verification, section quality rating, and collection sorting — and the code I got to delete.
AI StrategyFive Everyday Business Tasks a Decision API Can Handle
The non-technical guide to five back-office jobs a decision API quietly automates for small teams — support routing, content flagging, lead scoring, expense categorization, and form triage — mapped to Choose, Rate, and Verify.
AI StrategyWhy '70% Sure' Is More Useful Than 'Yes'
Decision models return a confidence score with every answer. Here's how to use it to automate the easy cases, flag the unclear ones, and build a cheap human-in-the-loop system without a data science team.
AI StrategyThe Three Questions a Decision Model Can Answer
Every decision an AI can make for you in software reduces to three shapes: choose from a list, rate on a scale, or verify yes-or-no. A plain-English guide to the three decision shapes, with examples and when to use each.
AI StrategyChatbot vs Decision Model: What's the Difference?
A chatbot writes you a paragraph; a decision model gives you a typed answer. A plain-English guide to how they differ, when to use each, and how the best AI products combine both.
AI StrategyWhat Is Jev? A Plain-English Explanation
Jev is TypeSafe AI's September 2026 decision model: feed it information and a question, get back a typed answer — not a paragraph. A plain-English guide to what Jev is, how it works, and whether it belongs in your stack.
AI StrategyNot Every AI Needs to Talk: Meet the Decision API
Chatbots get the headlines, but a quieter kind of AI — the Decision API — is doing the actual work inside modern software. What it is, when to reach for it instead of a chat completion, and the four patterns builders use most.
Fractional CTOWhy a Fractional CTO is the right call for SMBs in 2026
Hiring a full-time CTO too early burns runway. Hiring too late leaves you with sprawling architecture and accidental contracts. There's a middle path.
AI StrategyAn AI strategy for SMB owners that survives contact with reality
A pragmatic playbook for SMB owners deciding what to do (and not do) with AI in the next 12 months.
Zero-to-LaunchHow to ship a real product in 90 days when you've never done it before
A practical week-by-week framework for non-technical founders aiming for a real launch — not a demo — in three months.