Background

Latest news

Stay up to date with the latest developments and insights from Andri.ai

On Thinking Slowly: Why We Built Andri to Deliberate

On Thinking Slowly: Why We Built Andri to Deliberate

Most AI systems think once. We built something different. Not because it was fashionable, but because legal work demands the kind of reasoning that can't be rushed.

Read more →
Why Agentic Reasoning Is the Only Path to Production Legal AI

Why Agentic Reasoning Is the Only Path to Production Legal AI

If you're not building an agentic system—a loop of planning, tool use, reasoning and document generation grounded in a specific matter—you're not building production legal AI. You're building autocomplete with a legal skin.

Read more →
The In-House Triangle: How Andri Actually Works

The In-House Triangle: How Andri Actually Works

Your AI assistant forgot the defendant's address again. Here's how we make AI feel in-house: Personalisation sets the boundary. Tools make it real. Memory makes it repeat. The biological layers of DNA, education, and working memory map directly to model weights, training data, and context windows.

Read more →
Why legal AI needs to think, not just respond

Why legal AI needs to think, not just respond

The difference between useful legal AI and a fancy search box comes down to one thing: can it reason through problems the way lawyers do?

Read more →
Llama 4's 10M context window: bigger isn't always better for legal AI

Llama 4's 10M context window: bigger isn't always better for legal AI

Meta released Llama 4 with a 10 million token context window. Impressive engineering, but for legal work, what matters is finding the right information—not fitting everything in at once.

Read more →
Case simulation: see how your arguments hold up before you file

Case simulation: see how your arguments hold up before you file

We built a feature that simulates how opposing counsel and a judge might respond to your case. Here's what it does and why we think it matters.

Read more →
Why ChatGPT wrappers don't work for legal research

Why ChatGPT wrappers don't work for legal research

A lot of 'legal AI' is just ChatGPT with a different logo. Here's why that's a problem—and what a purpose-built system does differently.

Read more →
How we handle your documents (and why it matters)

How we handle your documents (and why it matters)

Legal documents need serious security. Here's exactly how Andri protects client files—encryption, isolation, access controls, and why we built it this way.

Read more →
OpenAI's Deep Research vs specialised legal AI

OpenAI's Deep Research vs specialised legal AI

OpenAI released Deep Research—an agent that trawls the internet for complex queries. Impressive tech, but fundamentally wrong for legal work. Here's why.

Read more →
Building Andri with Blokziel: what we learned from our first pilot

Building Andri with Blokziel: what we learned from our first pilot

Advocatenkantoor Blokziel became our first Dutch pilot firm. Here's what three months of working together taught us about what lawyers actually need from AI.

Read more →
Andri is now live in the UK

Andri is now live in the UK

We've spent the past months building proper UK legal coverage. Not just case law—the procedural details that actually matter: CPR rules, court forms, fee schedules, and the quirks of English and Welsh practice.

Read more →
EU regulations are now in Andri

EU regulations are now in Andri

We've added the full corpus of EU regulations, directives, and CJEU case law to Andri. Here's what that means for cross-border work and compliance matters.

Read more →
Why we're called Andri

Why we're called Andri

The name comes from the Library of Alexandria—the ancient world's attempt to gather all human knowledge in one place. Here's why that felt right for what we're building.

Read more →
First case won using Andri

First case won using Andri

Ronald Zwiers won his first case with Andri's help. Here's what we learned about building AI for actual legal work.

Read more →