Engineering Blog

Caching is all you need

Our cache hit rate looked great. Our cost data disagreed. Here's what we found when we stopped trusting the average.

How we built a lab to evaluate data agents

Inside Hex's eval architecture and the synthetic business it runs on.

How we built (and rebuilt) topic discovery into the Hex Context Agent

What are people actually asking your agent?

The golden age of the destination engineer

Why the engineers who care more about users than code are going to run the table.

We cloned our visualization team

They live inside a subagent now

Don’t be naked when the agents arrive

As AI agents absorb routine work, PMs and engineers need new leverage — judgment, focus, and clarity — to build the right thing.

Improving the Notebook Agent: error clustering

Why you should analyze your agent's reasoning, not just its errors

I'm sorry, but those are vanity evals

Using GPT-4.1 as a case study of our framework for impactful LLM evaluation

Beating browser bottlenecks with virtualization

Inside our new approach to virtualization

Making AI charts go brrrr

How we made AI charts generate 8x faster, going from an average response time of 10.3s to 1.3s

We took down production by misconfiguring our ETL

Here’s what we learned and how you can avoid it

Building a Builder

Designing and engineering a new app layout engine from scratch

A pragmatic approach to live collaboration

or: How I Learned to Stop Worrying and Love the Atomic Operation

On Software, Ships, and Shipping Incrementally

The only path around the 2.0 tar pit is incrementalism - replacing a legacy platform through small, deliberate steps.