From the Builders.
Notes on making AI the default, not the demo — from the people doing the work.
34 POSTS · RSS
What AI modernization actually is (and isn't)
Modernization is not a tool rollout or a pilot. It is three tracks — the software you build, the way your teams work, and the way the business runs — and the practice that keeps all three improving.
StrategyModernizationTracksSep 2, 2026 · 4 minWhat we mean by governed agents
An agent that works on launch day is easy. An agent that still works in month six, with someone accountable for it, is the actual product. Here is what governance means in practice.
AgentsGovernanceOperationsSep 2, 2026 · 4 minWhy pilots die in meetings
The AI pilot worked. Everyone was impressed. Nothing changed. Here is the mechanism behind that pattern, and what it takes to get from a demo to a default.
StrategyPilotsPracticeAug 18, 2026 · 3 minGuardrails as code
A policy an agent can exceed is a suggestion. The limits that matter live in the system: spend caps, action allow-lists, iteration limits, and the rule to ask instead of guess.
AgentsGovernanceAI EngineeringAug 4, 2026 · 3 minAI for companies that don't build software
Most of the businesses that would gain the most from AI have no engineering team and never will. The Operations track is for them. Here is what it looks like in practice.
OperationsModernizationSmall BusinessJul 21, 2026 · 3 minHow to tell a demo from a system
Every vendor has an impressive demo. Very few have a system. Seven questions that separate the two, and the answers that should end the meeting.
StrategyGovernanceCandidJul 7, 2026 · 3 minWhat AI work actually costs, and why we publish shapes instead of prices
A candid guide to where the money goes in an AI engagement, what drives it up or down, and why our packages list durations and cadences rather than numbers.
StrategyCandidJun 23, 2026 · 3 minWhat shipping a mobile platform in eight weeks taught us
We took a requirements document to the app stores in eight weeks. The product was not the interesting part. The delivery system we built first was.
EngineeringDeliveryCase NotesJun 16, 2026 · 3 minWhy one shipped workflow beats a training day
Teams forget most of what a training day teaches. They do not forget the thing they built. How we run buildcamps, and why the 30-day check-in is the part that matters.
DeliveryBuildcampEnablementJun 2, 2026 · 3 minRunning agents like operations
Once agents do real work, they are a process, and processes get managed. The metrics, the rhythms, and the roles that turn a pile of agents into an operation.
AgentsOperationsGovernanceMay 19, 2026 · 3 minFractional AI leadership or an advisory retainer: which one you need
Both put a senior practitioner alongside your leadership team. They answer different questions. How to tell which one your organization is actually asking.
StrategyLeadershipMay 5, 2026 · 3 minEvals in CI: a starter kit
The evaluation harness that lives in a notebook is one nobody runs. Here is the minimum to get AI behaviour tested on every pull request, without a platform.
EngineeringAI EngineeringEvalsApr 21, 2026 · 3 minWhat a good strategy readout contains
The output of a strategy sprint is not a deck. It is a decision, with the reasoning written down so it survives the next budget meeting. What we put in ours.
StrategyPracticeApr 7, 2026 · 3 minProve the idea in three weeks
Before betting a quarter on a modernization idea, build the small tool that shows whether it is real. Rapid tooling as a discovery method, not a cheap build.
EngineeringRapid ToolingModernizationMar 17, 2026 · 3 minThe two-person AI team
A small business does not need an AI department to run agents well. It needs two roles, clearly held, and a care plan behind them. Here is how that works.
OperationsAgentsSmall BusinessMar 3, 2026 · 3 minWhat an AI-ready codebase looks like
Making a codebase ready for AI does not mean rewriting it. It means making it legible, reachable, and testable, so both assistants and new capabilities can land safely.
EngineeringAI EngineeringSoftwareFeb 17, 2026 · 3 minWhat belongs in an agent's audit trail
When an agent does something wrong, the first question is what happened. Most agents cannot answer it. Here is what to record so yours can.
AgentsGovernanceFeb 3, 2026 · 3 minThe agent decay problem
Agents rarely crash. They slide. Success rates drift, people start correcting, then working around, and eventually nobody uses what it produces. How to see it coming.
AgentsGovernanceOperationsJan 20, 2026 · 3 minBuild, buy, or partner: a straight way to decide
Every AI initiative eventually hits the same fork. The answer is rarely the one the vendor pitch or the engineering team's pride suggests. A short decision method.
StrategyPracticeJan 6, 2026 · 3 minWhere AI pays off first
Most AI roadmaps start with the most exciting idea. Ours start with a scoring method. Here it is, with the four criteria and the trap each one avoids.
StrategyModernizationPracticeDec 16, 2025 · 3 minThe year AI stopped being a demo
A year-end note. The interesting change was not in the models. It was in what companies started asking for, and what they stopped accepting.
StrategyModernizationDec 2, 2025 · 3 minWhy we plan the handoff before we write code
A system that only works while the consultants are around is a subscription to consultants. Planning the exit first changes every decision that follows.
DeliveryPracticeAI EngineeringNov 18, 2025 · 3 minWhat a delivery audit actually looks at
Two weeks inside an engineering team, watching how work really flows. Here is what we look for, what we measure, and what the readout contains.
DeliveryPracticeNov 4, 2025 · 3 minSpec-driven development for people who don't write specs
AI-assisted engineering rewards teams that say what they want before they build it. That does not mean forty-page documents. It means a habit.
DeliveryAI EngineeringPracticeOct 21, 2025 · 3 minHuman gates: a field guide
Not every agent action needs a person's approval, and every agent action getting one is how automation dies. Where to put the gates, and where not to.
AgentsGovernancePracticeOct 7, 2025 · 3 minThe evaluation harness is the product
An AI feature without an eval is an opinion. Here is what an evaluation harness actually is, why we refuse to ship without one, and how to start small.
EngineeringAI EngineeringEvalsSep 5, 2025 · 6 minAugust 2025 AI Roundup
From Open Agents to Sonic Speed
IndustryAIMonthly RoundupAug 29, 2025 · 2 minLindy.ai 3.0 - From Idea to AI Co-Worker
Lindy.ai 3.0 empowers businesses with agent builder, autopilot and team collaboration.
IndustryAI ToolsApplied AIAug 6, 2025 · 7 minManaging Chaos
Why AI's Lightning Speed is Your Advantage
IndustryAIInnovationAug 5, 2025 · 5 minHow Small B2B Companies Are Outmaneuvering Corporate Giants
Examining how small B2B companies are leveraging artificial intelligence to outcompete larger corporations.
OperationsAIBusiness StrategyAug 5, 2025 · 4 minThe Small B2B Team's AI Playbook
7 AI Strategies That Let Small B2B Teams Punch Above Their Weight
OperationsAIB2BJul 15, 2025 · 2 minThe Future of Applied AI in Business
Exploring how artificial intelligence is reshaping modern business operations and creating new opportunities for growth.
IndustryAIBusiness StrategyJul 5, 2025 · 5 minTeam Labs: Accelerating Innovation Through Collaboration
Discover how innovation sprints and team workshops can transform your organization's capability to adapt and innovate.
DeliveryTeam DevelopmentInnovationJul 1, 2025 · 3 minBuilding Micro Software Solutions That Scale
Learn how to create lightweight, custom software solutions that grow with your business without the overhead of traditional platforms.
EngineeringSoftware DevelopmentMicro Solutions