How this actually gets done
Written for whoever has to implement it. Every guide carries a framework you can name, arithmetic you can rerun on your own numbers, and an honest account of where it goes wrong.
Getting started
How to Pick the First Process to Automate with AI
Most first automations are chosen by whoever complained loudest. Here is the selection method that survives contact wi
Read →11 min readThe AI Automation Readiness Check, Run in an Afternoon
A readiness questionnaire measures optimism. These five probes measure your systems. Run them in one afternoon and the
Read →12 min readWhy AI Pilots Fail: The Gap Between a Demo and Production
A pilot and a production system share a model and almost nothing else. Here are the five gaps that strand good demos,
Read →12 min readBuild, Buy or Wait: How to Choose Your AI Approach
Build versus buy is really six decisions, one per layer of the stack. Here is where to draw the line, what waiting act
Read →12 min readWriting an AI Automation Brief Your Vendor Can Price
A vendor cannot price what you have not decided. This is the brief that turns a vague automation idea into a fixed sco
Read →11 min readHow to Map a Process Before You Automate It
Swimlane diagrams do not stop automations from failing. What stops them is knowing the exception distribution and the
Read →11 min readThe Baseline You Must Take Before Any Automation
Without a before number, every after number is an opinion. Here are the five measurements to take, where the timestamp
Read →11 min readWhat Not to Automate, and How to Tell in Advance
Most automation failures were visible before the build started. Five disqualifiers, one test for silent errors, and th
Read →AI agents
What an AI Agent Actually Is, Minus the Hype
An agent is a loop. A model picks the next action from a set of tools, reads the result, and decides whether to keep g
Read →10 min readAgent or Workflow: How to Choose the Right Shape
Count the distinct step sequences in a hundred real cases. That number, not the demo you watched, tells you whether to
Read →11 min readHow to Design Tools an AI Agent Can Actually Use
Most reports of a model calling the wrong tool are really reports of a badly written tool description. Here is the con
Read →11 min readGiving an Agent Memory Without Giving It Amnesia
Agent memory is four stores with four different lifetimes, not one feature. Most memory bugs are a fact filed in the w
Read →12 min readWhen to Use Multiple Agents Instead of One Agent
A second agent buys you a boundary and charges you a handoff. Here is how to tell which side of that trade you are on,
Read →12 min readKeeping a Human in the Loop Where It Actually Matters
Approval on every step is not oversight, it is a queue with a rubber stamp at the end. This is how to spend a fixed am
Read →12 min readHow AI Agents Fail, and the Failure Modes to Design For
Agents rarely fail with a stack trace. They fail quietly, at the seams between the model and everything else, and the
Read →12 min readMaking an Agent's Decisions Auditable, Months Later
An audit trail is not a log file. It is the ability to reconstruct, a year later, exactly what the agent saw, which ru
Read →11 min readControlling What an AI Agent Is Allowed to Do
Agent permissions are an engineering problem, not a prompting one. Where to put the boundary, how to size the blast ra
Read →11 min readHow to Cost an AI Agent Before You Build It
Token price is the smallest line in an agent's bill. The full cost of a run, the loop arithmetic that surprises people
Read →Workflow automation
n8n vs Make vs Custom Code: How to Actually Choose
The tool argument is the wrong argument. What decides your stack is where the business logic lives, how often that log
Read →11 min readError Handling That Stops Silent Automation Failures
The automation that crashes is not your problem. The one that runs green every morning while quietly writing nothing i
Read →11 min readIdempotency in Automation: The Concept That Saves Your Data
Retries are not optional, so duplicates are not optional either unless you design them out. Idempotency is the propert
Read →12 min readRate Limits, Retries and Backoff, Explained Properly
Most retry code makes outages worse. It retries things that can never succeed, ignores the header telling it when to c
Read →11 min readShared Inbox Automation Without Losing a Single Email
A shared mailbox is a queue pretending to be a folder. Here is how to automate triage, routing and drafting without dr
Read →11 min readReplacing the Spreadsheet That Quietly Runs the Business
The file everyone depends on is four systems fused into one. Here is how to pull them apart, extract the rules nobody
Read →12 min readWebhooks vs Polling: How to Pick the Right Trigger
One gives you latency, the other gives you truth, and production systems need both. The delivery guarantees, the bound
Read →12 min readMigrating Off Brittle RPA Bots Without a Big Bang
Most RPA bots exist because somebody was refused an API. Here is how to inventory them, descend the stack rather than
Read →LLM applications and RAG
RAG Explained for People Who Have to Build It
The eight stages of a retrieval pipeline, the four gates every answer has to pass, cost arithmetic you can run on your
Read →12 min readChunking Strategies That Change Your Answers
How you split documents sets the ceiling on everything downstream. The five splitting methods, what each one destroys,
Read →12 min readWhy Your RAG System Gives Confident Wrong Answers
The six mechanisms behind wrong answers from a grounded system, why fluency is unrelated to evidence, the four conditi
Read →12 min readCitations and Grounding: Making Answers Checkable
Grounding and citation are two different engineering problems with two different tests. Here is what a citation has to
Read →12 min readWhen Fine-Tuning Beats Prompting, and When It Does Not
Fine-tuning teaches behaviour, not facts. The ladder to climb first, the one test that predicts whether tuning will he
Read →12 min readStructured Output: Getting JSON You Can Trust
Constrained decoding guarantees your JSON parses. It guarantees nothing about the values inside it. The schema design,
Read →12 min readPrompts as Code: Versioning, Testing, Shipping
A prompt is the most consequential configuration in an AI system and usually the least governed. What to version, why
Read →12 min readContext Windows: What Fits, and What Degrades
The advertised window is a ceiling the API enforces, not an amount the model uses well. What consumes the budget, why
Read →13 min readExtracting Data from Messy Documents Reliably
Most extraction failures are parsing failures wearing a model's clothes. The four document classes, the four gates eve
Read →Evaluation and reliability
Writing Evals for LLM Systems That Catch Real Failures
An eval is a frozen set of inputs, an assertion that decides pass or fail, and a number you can compare between versio
Read →12 min readBuilding a Golden Dataset That Does Not Rot in Six Months
The set of cases you measure against decides what you can see. Here is how to source it, stratify it, label it, and ke
Read →12 min readCatching Model Regressions Before Your Users Do
Four separate clocks can change your system's behaviour without anyone touching the code. Regression testing an LLM sy
Read →12 min readGuardrails That Do Not Break the Thing They Protect
A rule written in a prompt is a request. A rule enforced at the tool boundary is a control. Most of the cost of a guar
Read →12 min readMeasuring Whether an Automation Actually Worked
Most automation results are a before number and an after number with no control. Here is how to build a comparison tha
Read →13 min readMonitoring an AI System in Production Without Alert Noise
Uptime, latency and error rate can all be green while the system quietly answers wrongly. Here are the four planes wor
Read →13 min readWhat to Log in an AI System So You Can Debug It Later
A log is complete when you can replay a run without the original process. Here is the record to write for every model
Read →Running it
Rolling Out Automation Without a Team Revolt
A rollout is a deployment problem wearing a communications costume. Ship it in rungs, keep the old path alive, and let
Read →11 min readWho Owns the Automation After Launch: Ownership and Handover
Ownership is not a name on a wiki page. It is being paged, holding the access, and carrying the consequence, and those
Read →12 min readDocumentation That Survives the Person Who Wrote It
Most automation documentation rots because it describes state instead of decisions, and lives where no change ever for
Read →12 min readScaling from One Automation to Twenty: What Changes
The automations are not what breaks. Everything shared between them is: credentials, retries, alerts, the exception qu
Read →13 min readSecurity Questions to Answer Before You Ship AI
The five questions an AI feature adds to a security review you already run, the ladder that tells you which controls a
Read →12 min readHow to Handle Personal Data in an AI Pipeline
The engineering side of AI data privacy compliance: every place one record comes to rest, why redaction is risk reduct
Read →12 min readKeeping Humans Skilled When Machines Do the Work
Automation removes the easy cases, and the easy cases were the training ground. What decays, how fast, the arithmetic
Read →Costs and buying
What AI Automation Actually Costs to Build
Price is set by how many systems you touch and how accurate the output must be, not by the model. What drives a quote,
Read →13 min readToken Costs: The Arithmetic Nobody Shows You
Token cost is four multiplications and one division, and most estimates get the division wrong. The six meters a langu
Read →12 min readTotal Cost of Ownership for an AI System Over Three Years
Build and run are the two clocks everyone budgets. The two that break business cases are drift, which ticks on somebod
Read →12 min readIn-House vs Agency vs Freelancer for AI Development
Hourly rate is the least useful number in this decision. What each staffing model actually costs across eighteen month
Read →13 min readContract Terms That Actually Matter for AI Projects
A standard software statement of work assumes the deliverable either matches the spec or it does not. LLM systems are
Read →Want the system instead of the guide?
Send the process you want automated and we will scope the build.
Start a project →