For years, every claims feature at Alan meant building it twice. A year ago, we broke that cycle. Here's what that journey looked like — from expensive first steps to compounding returns, and AI's role in it all 👇. medium.com/alan/univers...

Universalisation at Alan: How We Stopped Building Everything Twice

From duplicate codebases to shared flows: how we turned multi-platform pain into compounding gains.

medium.com

Part 3 of our document processing series is out — how we scaled our pipeline from French insurance docs to three countries and multiple use cases, without rewriting everything. This wraps up the series. Thanks for following along! buff.ly/9cQL4F3

We think we built the best Python caching library in the world! RAM+Redis, atomic writes, distributed invalidation, async refresh, warmup, observability... Dive into this 5-part series, packed with diagrams, code snippets, dashboards, and numbers! medium.com/alan/alan-ca...

Few-shot examples are only as good as their labels. Ours were hallucinating values from memory. The LLM learned to do the same. Part 2 of the doc processing series: curated datasets, backtesting every change and separating parsing from human knowledge medium.com/alan/lessons...

Lessons from Running an LLM Document Processing Pipeline in Production

We evolved from text-only to multimodal, built curated reference datasets for RAG and developed an evaluation framework

medium.com

Ever wondered how Alan reimburses health claims so fast? Spoiler: we automated 70% of document processing. We just dropped the first article of our document processing series — stay tuned for more on how we're scaling this across other countries and use cases. buff.ly/2Lu7aWD

How we reached 70% document processing automation at Alan

We replaced our classic ML pipeline with a unified LLM-based approach using Markdown Transcription and few-shot RAG extraction

medium.com

📣 We’re doing something a bit different at @alan: an on-site Engineering Recruitment Day! 🗣️ Meet the team 👨‍💻 Technical and fit stages in 1 day 🚀 Clear checkpoint and quick next steps 📍 Paris (Alan HQ) 🗓️ Tue, March 17 ⏳ Apply by Wed, March 11 👉 jobs.ashbyhq.com/alan/a8b3fe4...

"Claude did it almost on its own while I was boiling water for my tea" That's how our engineers now fix flaky tests at Alan. Read more about our journey from manual debugging to AI-powered investigation in our latest article 👇 medium.com/alan/from-gu...

From Gut Feeling to AI assisted: How We Automated Flaky Test Investigation

Over the last year, we’ve improved our flaky tests investigation process, moving from manual debugging to an AI first solution. This…

medium.com

Don't chase leaderboards. There's always a score-latency tradeoff, and LLM providers often hide it. We reproduced Tau-bench to see for ourselves. Here's what we learned: medium.com/alan/benchma...

Benchmarking AI Agents: Stop Trusting Headline Scores, Start Measuring Trade-offs

Don’t chase leaderboards. Run the benchmark yourself and map your score–latency–cost frontier to choose models for production.

medium.com

There are many LLM benchmarks such as MMLU and GSM8k, but they're useless for AI agents. Real agents need to handle database state, tool calling, and multi-turn conversations. Stateful benchmarks show the path forward. New post on agent evaluation 👇

Benchmarking AI Agents: The Challenge of Real-World Evaluation

AI agents need stateful benchmarks. Unlike LLMs, agents interact with databases and users. We explore why and how to evaluate them…

medium.com

You've heard of the recent ISO27001:2022 certification of Alan by SGS, but want to know more about our journey towards certification? Head up to Maxime's post and enjoy the read! medium.com/alan/our-iso...

Our ISO 27001 journey: From security blueprint to certification success

Hey 👋 I’m Maxime, the ISMS lead at Alan, and I’d like to tell you about our ISO journey 🗺️

medium.com

Static chatbots couldn't handle complex support tickets about insurance claims. So we built something different with tool calls and the ReAct framework. Our Claim Agent investigates dynamically - just like human agents, but faster. Now automating 30% of tickets it receives.

After four years at Alan from Engineering to Security, I’ve seen what makes security succeed and here is what I learned: medium.com/alan/how-i-f...

How I found my security calling at Alan

This is the story of my journey from engineering to security at Alan, and what I’ve learned about building security that scales with our…

medium.com

🛠️ How we tamed the "works on my machine" chaos at Alan Engineering! Our new blog post reveals how Devbox transformed our dev experience, slashed onboarding time, and created consistent environments across our entire team. Check it out: medium.com/alan/from-ch...

From Chaos to Consistency: How Alan Transformed Developer Experience with Devbox

medium.com

In late January, DeepSeek shocked the world by dropping an open-weight successor to OpenAI's o1: R1. Their tech report discusses how to incentivize reasoning capability in LLMs. We share our learnings at:

DeepSeek R1: Demystifying LLM’s Reasoning Capabilities

DeepSeek shocked the world by dropping an open-weight successor to OpenAI’s o1: R1. This post summarizes our learnings from the tech…

medium.com

In 2024, we automated 20% of customer contacts with AI while maintaining the same customer satisfaction (the number is still growing!) 🚀 Learn about our journey:

How We Built Alan’s AI Assistant for Customer Support

At Alan, exceptional customer care isn’t just a service — it’s a core part of what sets us apart in the insurance industry.

medium.com