EngineersGovernance & Compliance
Technical readiness checklist for AI in production
Eight verifiable blocks, each with required evidence, to decide whether an AI system is ready for production — not just for a demo.
5 min read
Publication history
All articles
31 publications, ordered by date. Filter by category or tag to find what matters.
August 2026
July 2026
ExecutivesStrategy & ROI
The executive checklist: 30 questions before scaling AI
A verification routine across five dimensions — data, risk, integration, people and cost — to separate enthusiasm from readiness.
3 min read
EngineersData Architecture
Internal AI platform: from pilot to product
Golden paths, use case catalog, reusable components, environments, CI/CD, SLOs, and a support model for scaling AI beyond the pilot stage.
4 min read
ExecutivesStrategy & ROI
A 12-month roadmap: from isolated pilot to knowledge platform
Wave sequencing, exit criteria per phase, what to centralize and what to federate — and the mistakes that cost a year.
3 min read
EngineersGovernance & Compliance
Instrumenting cost and quality in AI production
How to measure tokens and cost per unit of work, attribute spend by team, monitor quality, and back your ROI calculation with real data.
4 min read
June 2026
ExecutivesStrategy & ROI
Enterprise AI ROI: measuring what is not on the invoice
Baseline, metrics per use case, efficiency gain versus accumulated asset — and how to avoid endless pilots that never reach production.
3 min read
EngineersApplied AI
Predictive maintenance: from sensor to the alert someone actually acts on
Why most predictive maintenance pilots don't survive past month six — and what changes when the alert is designed together with operations.
2 min read
EngineersData Architecture
Factory floor data collection architecture with MQTT and time series
From PLC to dashboard: topology, broker selection, topic modeling, and data retention for reliable industrial telemetry.
2 min read
ExecutivesStrategy & ROI
How to calculate the ROI of an industrial automation project
A practical method for turning gains in availability, quality, and labor into a number the investment committee will actually engage with.
2 min read
EngineersAI & Agents
Serving LLMs on-premise: GPU, quantization, and throughput
A technical guide to sizing GPU and memory, applying quantization, continuous batching, and KV cache when serving language models inside the company.
5 min read
May 2026
ExecutivesIntegration & Legacy
Cloud, private cloud or on-premise: where your AI should run
A decision matrix by data sensitivity and criticality: total cost, sovereignty, open versus proprietary models, and exit strategy.
3 min read
EngineersGovernance & Compliance
Immutable AI decision log with hash chaining
How to build an append-only, hash-chained record of AI decisions — with periodic anchoring, auditable correction, and vendor-independent export.
5 min read
ExecutivesGovernance & Compliance
An audit trail for decisions made with AI
What to record, how long to keep it, and how to answer an auditor, a customer or a regulator without depending on your vendor's goodwill.
3 min read
EngineersGovernance & Compliance
Policy as code for AI governance
How to turn your AI usage policy into executable rules: risk classification, attribute-based authorization, guardrails and CI tests.
4 min read
April 2026
ExecutivesGovernance & Compliance
The AI committee: who decides, who approves, who answers
Roles, a risk-based approval matrix, an acceptable use policy and a use-case inventory — the minimum executive governance for scaling without surprises.
3 min read
EngineersGovernance & Compliance
PII detection and masking in AI pipelines
How to detect, mask, and, when authorized, re-identify personal data in AI pipelines without breaking inference or data-subject rights.
5 min read
ExecutivesGovernance & Compliance
Data privacy in the age of AI: GDPR and LGPD in practice
Legal basis, minimization, anonymization, impact assessments and model-vendor clauses — what changes when AI enters the workflow.
3 min read
EngineersAI & Agents
Hybrid retrieval and per-user permissions in the index: how not to leak data in RAG
Lexical plus vector search with result fusion, source-inherited ACL applied at query time, revocation, multi-tenancy, and automated leakage tests.
5 min read
March 2026
EngineersAI & Agents
RAG engineering: chunking, embeddings, and reranking
Structure-aware chunking, embedding versioning, reindexing, reranking, and retrieval evaluation — the technical decisions that determine RAG quality.
5 min read
ExecutivesAI & Agents
RAG is not magic: the real cost of giving AI enterprise context
Answer quality is a data and permission problem, not a model problem. What changes when the index must respect who can see what.
3 min read
EngineersIntegration & Legacy
An anti-corruption layer to connect AI to a legacy ERP
API contracts, CDC vs. batch, idempotency, identity propagation, and load limits: the engineering of an anti-corruption layer between AI and core systems.
5 min read
ExecutivesIntegration & Legacy
Your ERP is 20 years old and AI needs to talk to it
Legacy integration patterns explained for the people who approve budgets — without rewriting the core or becoming hostage to a proprietary connector.
4 min read
EngineersGovernance & Compliance
Capture pipeline for AI interactions: from conversation to auditable knowledge
How to design the event schema, partitioning, retention, and indexing needed to turn prompts, retrieved context, and human edits into a reusable knowledge asset.
4 min read
February 2026
ExecutivesGovernance & Compliance
The forgotten asset: turning AI interactions into auditable knowledge
How to capture, version and curate prompts, responses and decisions so every AI use improves the next one — instead of vanishing at the end of the session.
4 min read
EngineersGovernance & Compliance
AI gateway: DLP and outbound data control
How a corporate AI gateway becomes the single point of egress — identity, routing, DLP on request and response, cost limits and shadow AI blocking.
6 min read
ExecutivesGovernance & Compliance
Where what your employees type into AI ends up
The real path of a corporate prompt: retention, model training, subprocessors and data residency — and the questions that belong in the contract.
4 min read
EngineersAI & Agents
Continuous LLM evaluation: regression datasets that survive model swaps
How to build a regression dataset from real cases, pick metrics that matter, and use an LLM-as-judge calibrated against human review to block bad deploys in CI.
5 min read
January 2026
EngineersAI & Agents
Prompts as code: versioning, testing, and evaluation
Prompts and system instructions treated with the same discipline as code: repository, PR review, automated tests, canary rollout, and rollback.
4 min read
ExecutivesAI & Agents
Prompt engineering didn't die. It moved up a layer.
What actually replaces the per-turn prompt in agentic systems — and where the risk goes when no one reviews the wording anymore.
9 min read
EngineersAI & Agents
Reference architecture for enterprise AI: layers, contracts, and perimeter
Model gateway, orchestration, retrieval, memory, identity, observability, and audit trail: the technical design behind an enterprise AI that isn't locked to a single vendor.
4 min read
ExecutivesStrategy & ROI
Your company's ChatGPT is not an AI strategy
Individual licenses create personal gain and organizational loss. What a company forfeits when the knowledge produced with AI dies inside the chat window.
4 min read
Put it to work
Bring this into your company with the e.works platform
Everything we cover here — AI governance, legacy system integration, audit trails and preserved corporate knowledge — is available on the e.works platform at eworks.cloud.
e.works infrastructure on AWS
A managed environment protected by e.works on AWS, with encryption, per-company isolation, backup and high availability.
On-premises, in your environment
The same platform running in your company's data center or private cloud, when data sovereignty requires that nothing leaves your perimeter.
In either model your data stays yours — with access control, audit logging, configurable retention and guaranteed availability.
Newsletter
Technical and strategic content, once a month
Analysis on automation, industrial data and technology adoption. No spam.