Skip to main content
Insights

Field notes from the engineering floor.

Practical articles written by engineers, for people who make technical decisions. No growth hacks. Occasional jokes. Mostly useful.

Engineering guidance, not product guarantees. Verify vendor details before deciding.

All articles

11 of 11 articles

ERP & integrations

3
  1. NewIntegrationIdempotency, explained with a cup of coffee.Why retried API calls create duplicate orders, and how idempotency keys make integrations safe, explained with a coffee order.8 min read
  2. OdooHow to plan an ERP migration without guesswork.A practical framework for inventorying customizations, validating data, rehearsing cutover, and communicating an ERP migration plan.8 min read
  3. OdooWhy most Odoo implementations fail, and how to avoid it.ERP failures are rarely technical. The organizational patterns behind failed Odoo projects, and the process changes that prevent them.9 min read

AI in production

3
  1. NewAIRAG vs fine-tuning: an open-book exam or a semester of study?When to use retrieval-augmented generation, when to fine-tune, and when to just write a better prompt. A practical comparison.8 min read
  2. AIDesigning human review into document AI.Design extraction workflows with validation rules, confidence handling, exception queues, and a path for reviewer corrections.8 min read
  3. AIBuilding production AI agents: what the tutorials don't tell you.LLM demos are easy; production AI is hard. Structured outputs, observability, human review, and adversarial testing for real agents.8 min read

Cloud & data

3
  1. NewDatabasesPostgreSQL indexes without tears.A friendly guide to PostgreSQL indexes: B-tree, composite, partial, and GIN indexes, reading EXPLAIN, and when indexes hurt.8 min read
  2. InfrastructureDo you actually need Kubernetes? An honest flowchart in prose.An honest decision guide: when Kubernetes is worth its complexity, and when a VM, Docker Compose, or a managed platform is the better call.8 min read
  3. InfrastructureThe architecture behind high-performance Odoo deployments.How large Odoo deployments stay fast: worker processes, load balancing, PostgreSQL tuning, replicas, caching, and long-polling.8 min read

Apps & engineering

2
  1. EngineeringTechnical debt is a credit card, not a crime.Technical debt explained for non-engineers: when borrowing against the future is smart, how interest accrues, and how to pay it down.8 min read
  2. MobileOffline-first sync: decide what happens when the network returns.A design checklist for local writes, queued operations, conflict resolution, retries, and user feedback in offline-capable apps.8 min read

What we write about

These articles come from the problems we meet in real projects: ERP implementations that drift, integrations that create duplicates, databases that slow down as they grow, AI systems that work in a demo and struggle in production, and infrastructure decisions that are harder to undo than to make.

We write for the people who have to live with the result: operations managers, finance leads, CTOs, and the engineers who maintain the system after launch. Each piece aims to leave you with a practical decision framework, a checklist, or a pattern you can apply the same week.

Topics and where to start

If you're working onStart with
An Odoo rollout or upgradeWhy Odoo implementations fail, and How to plan an ERP migration
Odoo performanceHigh-performance Odoo architecture, and PostgreSQL indexes without tears
AI in productionBuilding production AI agents, and RAG vs fine-tuning
Document automationDesigning human-reviewed document AI
IntegrationsIdempotency explained with coffee
Mobile apps in the fieldOffline-first sync
Infrastructure choicesDo you actually need Kubernetes?
Codebase healthTechnical debt is a credit card

How we write

We avoid hype, invented statistics, and vendor talking points. When an approach depends on your situation, we say so and explain what it depends on. If you disagree with something, or want us to cover a topic, tell us through the contact page; several of these articles started as questions from clients.

Reading paths by role

  1. Operations or finance lead

    Start with why Odoo implementations fail, then how to plan an ERP migration.

  2. CTO or engineering lead

    Start with technical debt as a credit card, then do you need Kubernetes.

  3. AI or innovation lead

    Start with RAG vs fine-tuning, then building production AI agents.

  4. Developer or architect

    Start with idempotency, then PostgreSQL indexes and offline-first sync.

How the articles are structured

Each article explains the problem, the main options with their trade-offs, the mistakes we see most often, and a short checklist or set of takeaways you can use straight away. Code examples are kept small and readable, and we link to our service pages only where they genuinely help you go further.

Corrections and updates

Each article shows when it was published, and we revise articles when tools change or our experience teaches us something new. If you find an error or something that no longer holds, let us know through the contact page and we'll fix it.

What we don't write about

You won't find trend roundups, tool listicles, or predictions about the next big thing here. We write about problems we have solved, or watched others struggle with, in real systems: the details that decide whether a project works in production, not just in a demo.

We also avoid publishing benchmark numbers we can't reproduce for you. Where performance matters, we explain how to measure it on your own system instead, because your data, workload, and hardware are what really decide the result.

Stay in touch

New articles appear here first. Our monthly newsletter, from the signup in the site footer, summarizes what's new along with the occasional lesson from recent projects.

How we check the facts

Every statistic in an article links to its source, and we only cite figures we've read at the source rather than second-hand. When a vendor detail changes, such as a support period or a pricing model, we update the article and the review date at the bottom.

Disagree with something?

Good. Engineering advice depends on context, and yours may differ from the projects we draw on. If you think an article is wrong or missing an important case, tell us through the contact form. If you're right, we'll update it and say so.

Frequently asked questions

Who writes these articles?
Anvexis Stack, based on the problems we see in real projects. Each article lists its sources so you can check the facts.
Can we republish an article?
Please link to it rather than copying it. That way readers always see the latest, corrected version.
Can you write about a topic we care about?
Suggest it through the contact form. If it's a question many teams face, it's a good candidate for an article.