Blog

AI-assisted SDLC in a regulated industry: case study

tomasz.grynkiewicz@netguru.com (Tomasz Grynkiewicz) — Mon, 02 Nov 2026 10:26:57 GMT

We run an AI-assisted SDLC on a multi-tenant SaaS platform for a client in a regulated industry, with Claude Code writing code, requirements, and tests. The agent has no permission a junior engineer would not have: three levels of CI gates and five human decision points tie every change to a requirement, a test, and a release.

Is your ecommerce platform holding back growth? 7 signs it's time to switch

kacper.rafalski@netguru.com (Kacper Rafalski) — Fri, 30 Oct 2026 11:52:31 GMT

When the board asks for a replatform, the first question is whether the platform is the cause at all. Think of a warehouse: messy shelving, too few staff and badly sequenced pick routes are clutter problems, fixable by reorganising and certain to follow you into any new building.

DevOps transformation: Operating model, roadmap and metrics

kacper.rafalski@netguru.com (Kacper Rafalski) — Wed, 28 Oct 2026 17:42:41 GMT

A DevOps transformation is an operating-model change: teams own software from commit to production, and delivery is measured by outcomes, not tool adoption. Pipelines, infrastructure as code and platforms are the means, not the goal.

Data readiness assessment: checklist, scoring and how to run one

kacper.rafalski@netguru.com (Kacper Rafalski) — Mon, 26 Oct 2026 11:00:01 GMT

A data readiness assessment answers one question: does the data behind a specific use case exist, can it be used legally and technically, and is it good enough? Readiness is a property of the pairing between data and use case, so the same warehouse can be ready for a dashboard and unfit for a RAG assistant.

What is security architecture? Components, frameworks and design steps

kacper.rafalski@netguru.com (Kacper Rafalski) — Fri, 23 Oct 2026 07:30:00 GMT

Security architecture is the design layer that decides where trust changes, which controls guard each boundary and which risk or requirement each control answers. A set of purchased tools is not an architecture, because tools without that traceability leave gaps nobody owns.

What is technology consulting? Services, types and when to hire (2026)

kacper.rafalski@netguru.com (Kacper Rafalski) — Wed, 21 Oct 2026 09:15:00 GMT

Technology consulting is paid, independent advice on which technology a company should choose, how to structure it and in what order to build it, ending in documents a team can act on. The consultant owns the recommendation, not the running system.

Akeneo vs Bluestone PIM for B2B: architecture, cost and fit

kacper.rafalski@netguru.com (Kacper Rafalski) — Mon, 19 Oct 2026 09:15:00 GMT

Akeneo and Bluestone PIM solve the same problem with different assumptions about who does the engineering. Akeneo offers a large collection of connectors and a choice between a self-hosted Community Edition and the SaaS Akeneo Product Cloud.

Data governance implementation: A 30/60/90 day playbook

kacper.rafalski@netguru.com (Kacper Rafalski) — Mon, 19 Oct 2026 06:15:00 GMT

Data governance implementation rarely fails for lack of a framework. It fails in the months after kickoff, when policy documents pile up faster than anyone enforces them. What decides the outcome is sequencing: which decision rights you assign first, which datasets you govern first, and which tools you hold back until the organization is ready to use them.

Internet of Everything (IoE): What it is and how it differs from IoT

kacper.rafalski@netguru.com (Kacper Rafalski) — Fri, 16 Oct 2026 07:00:00 GMT

Internet of Everything (IoE) gets treated as a rebrand of IoT, but the distinction matters at the architecture level. IoT is a connectivity layer; IoE is an operating model that binds that layer to people, process, and decision workflows.

Hexagonal architecture: Ports and adapters guide for devs

kacper.rafalski@netguru.com (Kacper Rafalski) — Wed, 14 Oct 2026 06:15:00 GMT

Hexagonal architecture keeps business rules in a domain core and pushes every database, API and framework out to replaceable adapters. The pattern isn't about hexagons. It's about where you draw the boundary between business rules and everything else, and how disciplined you stay about keeping infrastructure on the far side of it.

Cross-functional collaboration: A framework that works

kacper.rafalski@netguru.com (Kacper Rafalski) — Mon, 12 Oct 2026 06:45:00 GMT

Cross-functional collaboration breaks down not because people won't cooperate, but because no one owns the decision. Teams meet, share updates, align on OKRs, and still stall at the moment someone needs to say yes or no. For engineering leaders managing 50-500 person orgs, the fix isn't more meetings; it's a decision rights framework that names who decides, who's consulted, and who just needs to know.

Design thinking process: The 5 stages explained for teams

kacper.rafalski@netguru.com (Kacper Rafalski) — Fri, 09 Oct 2026 08:15:01 GMT

Most teams treat design thinking as a poster on the wall, not a workflow: they run one ideation workshop and call it done, then wonder why the shipped feature misses the mark. The real value comes from treating Empathize, Define, Ideate, Prototype, and Test as a loop that runs alongside sprint planning, not a phase that precedes it.

What does a product designer do? Role, skills & salary

kacper.rafalski@netguru.com (Kacper Rafalski) — Wed, 07 Oct 2026 08:45:00 GMT

Job listings for "product designer," "UX designer," and "UI designer" often describe the same job with three different titles, and that inconsistency is exactly why the role confuses both job seekers and hiring managers. A product designer owns the full arc from user research to shipped interface, working across product, engineering, and business goals rather than just visuals or flows in isolation.

What is dbt in data engineering? Architecture guide

kacper.rafalski@netguru.com (Kacper Rafalski) — Mon, 05 Oct 2026 10:30:00 GMT

dbt (data build tool) isn't a pipeline orchestrator or a warehouse, it's the transformation layer that turns raw loaded tables into tested, documented, version-controlled models using SQL and a compiler. Most teams misjudge where it sits, bolting it onto Airflow or treating it like ETL software, then wonder why lineage breaks.

Modern data stack: Layers, tools, and design guide

kacper.rafalski@netguru.com (Kacper Rafalski) — Fri, 02 Oct 2026 10:15:00 GMT

A modern data stack fails most often not from a bad tool choice, but from bolting cloud-native components onto a batch-era mental model. Teams swap Redshift for Snowflake, add dbt, and still wait days for trusted metrics because the semantic layer, observability, and governance pieces never made it into the design.

AI readiness assessment: Framework, checklist & scoring model

kacper.rafalski@netguru.com (Kacper Rafalski) — Wed, 30 Sep 2026 09:00:00 GMT

Most AI initiatives don't fail because of the model, they fail because the organization wasn't ready for it: fragmented data, no governance owner, and teams without AI literacy. An AI readiness assessment exists to surface those gaps before budget is committed, not after a pilot stalls.

Usability testing: A practitioner's guide to running it right

kacper.rafalski@netguru.com (Kacper Rafalski) — Mon, 28 Sep 2026 07:30:00 GMT

Usability testing fails most often not because teams skip it, but because they run it without a facilitator plan, a clear task script, or a way to score results. A small, well-scripted test with five participants will tell you more than a large study with no plan.

Headless CMS migration without breaking the content model

kacper.rafalski@netguru.com (Kacper Rafalski) — Thu, 24 Sep 2026 10:30:00 GMT

A headless CMS migration is not a content transfer, it's a schema translation between two incompatible data models, and treating it as a copy-paste job is why so many migrations stall mid-flight or get quietly rebuilt six months in. The failures aren't exotic: they're field-type mismatches, flattened relationships that need re-modelling as references, locale structures that don't map, and asset URLs baked into rich text.

The strangler fig pattern: modernizing without a big-bang rewrite

kacper.rafalski@netguru.com (Kacper Rafalski) — Tue, 22 Sep 2026 09:30:01 GMT

A strangler fig migration doesn't fail on day one. It fails near the end, when the remaining slices turn out to be unscoped and the business has already banked the win.

Application portfolio rationalization: deciding what to modernize first

kacper.rafalski@netguru.com (Kacper Rafalski) — Thu, 17 Sep 2026 09:30:00 GMT

Application portfolio rationalization gets treated as a scoring exercise: rank every application, plot it on a matrix, retire the bottom quartile. That's the easy 20%.

Legacy system integration: when to wrap a system instead of replacing it

kacper.rafalski@netguru.com (Kacper Rafalski) — Tue, 15 Sep 2026 09:00:00 GMT

Wrapping a legacy system and modernizing it solve different problems, and treating them as interchangeable is what stalls most integration efforts. The real question isn't 'can we expose this data', that's almost always solvable with an integration layer.

Boosting Engagement with AI: Personalized Marketing Emails & Promotions

kacper.rafalski@netguru.com (Kacper Rafalski) — Mon, 14 Sep 2026 08:21:29 GMT

As companies strive to enhance user engagement, leveraging AI to personalize email campaigns and promotional materials is becoming an increasingly popular solution.

What is a legacy system? 12 signs yours has become one

kacper.rafalski@netguru.com (Kacper Rafalski) — Thu, 10 Sep 2026 08:15:00 GMT

A legacy system isn't defined by age, a five-year-old platform can be legacy while a twenty-year-old mainframe isn't. The real marker is a capability-need gap: the system can no longer do what the business now needs, safely or affordably.

How to choose a legacy modernization partner (and the questions that expose a bad one)

kacper.rafalski@netguru.com (Kacper Rafalski) — Tue, 08 Sep 2026 07:45:01 GMT

Every list of application modernization companies is either an agency's self-ranking or a vendor-authored roundup with a sponsor buried in the footer. Neither gives an enterprise buyer a way to actually differentiate one vendor from another.

What is application modernization? Definition & 2026 guide

kacper.rafalski@netguru.com (Kacper Rafalski) — Fri, 04 Sep 2026 07:15:00 GMT

Application modernization is the practice of updating an existing application's architecture, platform, or code so it keeps meeting current business demands, instead of replacing it outright. It preserves the business logic an organization already trusts and replaces the parts of the stack that can no longer keep pace.

AI voice agent for customer service: Where it works today

kacper.rafalski@netguru.com (Kacper Rafalski) — Thu, 03 Sep 2026 11:15:00 GMT

Most AI voice agent pilots fail not because the model can't talk, but because teams deploy it where chat or IVR would've done fine, or skip the compliance groundwork a vertical actually requires.

Real-time voice AI low-latency techniques that work

kacper.rafalski@netguru.com (Kacper Rafalski) — Tue, 01 Sep 2026 07:45:00 GMT

Sub-second voice AI isn't a model problem, it's a pipeline problem. Most teams optimize the LLM and ignore the 400ms bleeding out of STT finalization, TTS buffering, and network jitter.

Financial compliance software: Evaluation guide for 2026

kacper.rafalski@netguru.com (Kacper Rafalski) — Thu, 27 Aug 2026 07:00:02 GMT

Most financial compliance software evaluations fail for the same reason: buyers treat GRC platforms, transaction monitoring systems, and obligation management tools as interchangeable when they solve different problems. A compliance officer shortlisting vendors for AML and KYC coverage needs different architecture than one managing GDPR data-subject requests or Dodd-Frank reporting.

How Allegro Builds AI-First Design Systems for Predictable Scale

kacper.rafalski@netguru.com (Kacper Rafalski) — Wed, 26 Aug 2026 09:29:50 GMT

AI is rewriting product craft. Marcin Łasica, Head of Design Excellence at Allegro, shares how his team builds for speed, quality, and scale on the Next in Commerce session.

Software architecture patterns: Full catalog & how to choose

kacper.rafalski@netguru.com (Kacper Rafalski) — Tue, 25 Aug 2026 06:59:01 GMT

Choosing a software architecture pattern is rarely about finding the 'best' one, it's about matching trade-offs to constraints most teams never state explicitly: deployment cadence, team topology, and failure tolerance. Confusing an architecture style (a broad philosophy, like distributed systems) with a pattern (a concrete, named approach, like CQRS) leads teams to bolt microservices onto problems that layered monoliths solve better.

Stage-gate process: How gates and stages drive NPD

kacper.rafalski@netguru.com (Kacper Rafalski) — Fri, 21 Aug 2026 06:47:25 GMT

Most stage-gate failures aren't caused by a weak framework, they're caused by gatekeepers who treat go/kill/hold reviews as scheduling checkpoints instead of real decisions. Robert G. Cooper's model was built to force disciplined kill decisions early, before capital-intensive development spend, yet most teams water it down until every gate is a rubber stamp.

Latest AI developments 2026: Models, agentic AI, infrastructure

kacper.rafalski@netguru.com (Kacper Rafalski) — Wed, 19 Aug 2026 08:45:00 GMT

Most 'AI news' roundups confuse volume with signal, a dozen model releases a week, but only one or two that change a build-versus-buy decision. For engineering leaders, the real question isn't what launched, it's what shifts your inference cost, your vendor lock-in risk, or your agentic deployment timeline.

Agentic AI development services: Build, buy, or partner guide

kacper.rafalski@netguru.com (Kacper Rafalski) — Fri, 14 Aug 2026 07:45:00 GMT

Most agentic AI initiatives stall not because the technology fails, but because enterprises skip the scoping decision entirely, jumping into vendor calls before answering whether this should be built in-house, outsourced, or run as a partnership. That single decision determines your architecture, your governance burden, and your total cost of ownership for years.

Data governance services: Building an AI-ready framework

kacper.rafalski@netguru.com (Kacper Rafalski) — Wed, 12 Aug 2026 06:53:13 GMT

Data governance programs fail most often not from weak policy but from being designed for compliance audits instead of AI consumption. Enterprises that treat governance as a checkbox exercise end up with clean reports and unusable training data.

Mobile application security testing: Tools & methods guide

kacper.rafalski@netguru.com (Kacper Rafalski) — Thu, 06 Aug 2026 11:25:12 GMT

Most mobile app security programs fail not because teams skip testing, but because they run a single scan and call it coverage. Effective testing combines static and dynamic analysis mapped to OWASP MASVS controls. This guide breaks down what each method catches, what it misses, and how to sequence the right tools before shipping.

AI, Sustainability & Grocery: How Smart Commerce Can Reduce Waste and Carbon Footprint

Krystian Bergmann — Thu, 06 Aug 2026 09:31:32 GMT

Every supermarket runs the same quiet ritual. Late in the afternoon, someone walks the aisles with a sticker gun, hunting for the yogurt that expires tomorrow and the bread that won’t survive the night.

Website maintenance for custom-built sites: 2026 guide

kacper.rafalski@netguru.com (Kacper Rafalski) — Tue, 04 Aug 2026 07:52:34 GMT

Website maintenance for a custom-built application isn't a WordPress plugin-update checklist stretched to fit, it's a different discipline entirely. Most teams that skip it don't notice the failure until a dependency chain silently breaks a build, a framework hits end-of-life mid-project, or Interaction to Next Paint quietly drifts past the threshold that was fine a year ago.

Flexible Fulfillment Strategies for 2026: BOPIS, Click-&-Collect, Dark Stores & Micro-Fulfillment Centers

barbara.rybicka@netguru.com (Barbara Rybicka) — Mon, 03 Aug 2026 13:59:55 GMT

In November 2025, Kroger took a $2.6 billion write-down and announced it was closing three of the most sophisticated grocery warehouses ever built in America. The replacement plan was a store employee with a cart.

RAG chatbot accuracy: how data preparation beat model size

Grzegorz Skrzypek — Tue, 28 Jul 2026 10:10:57 GMT

You would think a bigger model means fewer hallucinations. More parameters, more reasoning tokens, more accuracy. Well, we hit 98% accuracy and near-zero hallucinations with a very small model instead. How?

Agentic AI: Turning Complexity Into Clarity for the C-Suite

kacper.rafalski@netguru.com (Kacper Rafalski) — Mon, 27 Jul 2026 14:30:40 GMT

AI is rewriting the playbook in real time. Joann Wu, VP and Global Head of Product Design at Uber, shares how design leaders can adapt in our Next in Commerce session.

Win the Zero‑Click Era: Decathlon’s AI Commerce Playbook for ROI

kacper.rafalski@netguru.com (Kacper Rafalski) — Mon, 27 Jul 2026 14:24:51 GMT

Retail is entering the zero‑click era. Aleksander Drupkowski, Director of E‑Commerce Operations and Development for Central Eastern Europe at Decathlon, shares how to win in our AI Commerce at Decathlon: Automation, Agents & the Zero‑Click Era session.

Scale AI In Wind: Process-First Playbook For Faster Decisions

kacper.rafalski@netguru.com (Kacper Rafalski) — Mon, 27 Jul 2026 14:19:25 GMT

AI won’t fix broken processes—leaders will. Rodrigo Álvarez-Bandrés, Enterprise Process Innovation Program Manager at Siemens Gamesa, shares a pragmatic playbook for scaling AI from the Next in Commerce session.

Eric Ries on AI: Replace Speed With Validated Learning for Growth

kacper.rafalski@netguru.com (Kacper Rafalski) — Mon, 27 Jul 2026 14:10:17 GMT

In the AI rush, speed without learning is a trap. Eric Ries — author of The Lean Startup and founder of Long Term Stock Exchange and Answer.AI — shares pragmatic guidance in this Next in Commerce session.

AI Integration vs Adoption Speed in B2B: What Actually Wins

barbara.rybicka@netguru.com (Barbara Rybicka) — Wed, 22 Jul 2026 07:38:46 GMT

Your board wants to hear you shipped AI. That's the wrong milestone. The better question — the one this article argues every AI roadmap should be built around — is: which business decisions are now different because of AI?

AI in social media: Feed ranking, content & chatbot tools

kacper.rafalski@netguru.com (Kacper Rafalski) — Fri, 17 Jul 2026 08:00:00 GMT

AI already runs underneath every major social platform: ranking models decide what each user sees, generative tools draft captions and creatives, predictive models assemble audiences, and chatbots qualify leads before a sales rep ever logs in. For marketing and product teams, the question is no longer whether to use AI in social media — it's which layers of that stack to adopt deliberately.

Best AI development companies in 2026: 15+ ranked

kacper.rafalski@netguru.com (Kacper Rafalski) — Wed, 15 Jul 2026 10:30:01 GMT

By Q2 2026, most engineering leaders aren't asking whether to build with AI, they're asking which external partner can compress a 12-month ML roadmap into a single quarter without creating a model-drift liability they'll inherit forever. The gap between a vendor who wires LangChain to an OpenAI endpoint and one who designs production-grade RAG pipelines, owns MLOps instrumentation, and ships ISO 27001-audited infrastructure is not visible on a agency website.

VTEX composable commerce: Pragmatic Composability explained

kacper.rafalski@netguru.com (Kacper Rafalski) — Mon, 13 Jul 2026 07:00:01 GMT

VTEX takes a third position on composability, Pragmatic Composability, and it's generated serious debate among platform architects.

Embedded analytics: architecture, integration, and build vs. buy

kacper.rafalski@netguru.com (Kacper Rafalski) — Fri, 10 Jul 2026 10:30:01 GMT

Your product ships. Users log in, do their work, then export CSVs to Excel to answer the question your app almost answered.

VTEX omnichannel: ship-from-store, pickup & endless aisle

kacper.rafalski@netguru.com (Kacper Rafalski) — Fri, 10 Jul 2026 09:00:01 GMT

Most VTEX omnichannel projects stall not at the vision stage but at the configuration layer, when a retail ops team discovers that Ship-from-Store, click-and-collect, and Endless Aisle each depend on the same upstream prerequisite: unified, real-time inventory visibility across every node. Get that foundation wrong and your 50km pickup radius returns no eligible stores, your Endless Aisle app surfaces out-of-stock SKUs, and your OMS routes orders to the slowest warehouse by default.

AI Chatbot Evaluation and Observability with Promptfoo and Langfuse

Krystian Bergmann — Thu, 09 Jul 2026 07:57:31 GMT

A chatbot that passes demo questions can still fail the moment real users take over. Here's how to evaluate chatbots before release and monitor them after, using Promptfoo for testing and Langfuse for observability.