16 articles on industry insights from the Alongside engineering team.
Portugal’s AMÁLIA LLM is a strong step toward European Portuguese AI. The bigger lesson: useful national AI needs open models, local data, and better evaluation.
Everything Portuguese real estate agents (mediadores imobiliários) need to know about Law 83/2017 and AML obligations before the IMPIC: identification, suspicious operation reporting, and penalties.
Pinecone and Modal are not direct substitutes. One solves retrieval. The other solves execution. Strong AI architecture starts by separating those decisions.
Technical due diligence should reveal delivery, security, and maintainability risk before commercial commitments harden, not after the deal is already operational.
The Cyber Resilience Act puts more pressure on software producers to prove security by design, vulnerability handling, and post-market discipline.
A good NIS2 gap analysis does not produce a generic control list. It shows which governance, security, and resilience gaps create the most exposure right now.
DORA changes the sales conversation for vendors selling into financial services. Buyers increasingly want evidence of resilience, incident readiness, and supplier discipline.
The AI Act is not just a legal issue. Product teams need clearer risk classification, stronger documentation, and better release controls long before enforcement pressure peaks.
Dedicated teams and staff augmentation solve different problems. The right choice depends on how much ownership, speed, continuity, and product context your delivery challenge really needs.
Cybersecurity advisory support is most valuable before an incident, audit failure, or stalled transformation forces a rushed decision. Here is when outside guidance creates the biggest payoff.
Many organisations treat NIS2 as a paperwork exercise and miss the operational gaps that regulators and attackers both notice first. Here are the readiness mistakes that create the most exposure.
A strong cybersecurity audit should do more than confirm control existence. It should show where risk is actually concentrated and what remediation will reduce exposure fastest.
Security reviews create value when they shape scope, architecture, and rollout decisions early. As a late gate, they mostly create surprise and rework.
It's happening again. Just as you get comfortable with one software tool, another one launches and suddenly everyone's wondering whether they need to switch. This week, we watched OpenAI drop ChatGPT 5.5 (literally today), and a week before that, Anthropic released Claude Opus 4.7. So which one should you care about? Here's the thing: they're both incredible, but they're built for different things. Let me break this down without all the technical nonsense.
Yann LeCun — Turing Award winner, ex-head of Meta AI, one of deep learning's original architects — walked away from Meta and raised $1.03 billion for a new company called AMI Labs. It's Europe's largest seed round ever. His bet: the entire LLM paradigm is wrong, and real machine intelligence needs a completely different foundation.
What building the future told us in the first day?
