Blog | Dweve Engineering Updates

Technical essays, briefs, and analysis from the team.

Every release, every change, every upgrade path. Monthly stable, 3-year LTS.

SDKs, integration guides, and architecture documentation.

Technical papers and sector briefings. All open-access, all reproducible.

Every claim in every post is reproducible. The code is open. The benchmarks are in the repository. If a number looks wrong, run it yourself and tell us.

No account, no gated posts, no tracking scripts.

Engineering posts include specific design decisions, rejected alternatives, benchmark tables, and the honest trade-offs that produced the final implementation.

Follow compiler internals, model routing, binary retrieval indexes, attention kernels, and the edges where implementation is moving faster than the documentation. The engineers doing the work write for readers who want the design details.

The RSS feed is the canonical source. Email updates go out once per month summarising new posts; we do not send promotional email, and we do not share the list with anyone for any reason.

Subscribe for the signal, skip the hype.

Policy and essays land when the question earns the room.

No preliminary results that might mislead.

Notes written by the engineer who merged the change.

We do not publish filler to hit a cadence. Quiet months mean the team is head-down shipping. Busy months mean something worth documenting has happened. Our average is approximately two posts per month, but the variance is high. A major release can produce several posts in a week. A deep research phase can produce nothing for six weeks.

Each post has a claim, evidence, and a conclusion you can cite. Not opinions. Not a listicle. Just writing from the people who did the work.

Release notes and engineering accounts of what changed, what broke, and what the team learned. Written by the people who did the work, with enough context to cite in a procurement review.

Subscribe to the RSS feed or the monthly email digest. No tracking, no ads, no spam. Just writing from the people building the technology.

Customers and contributors writing in their own voice about what they did with Dweve. We do not edit them into our voice; we let them sound like themselves.

A short post, a half-formed thought, an opinion that is still warm. Sometimes a comment on someone else's essay; sometimes a teaser for upcoming work. We do not pretend every short post is a research paper.

Once or twice a quarter, someone on the team writes a longer piece. Why Europe needs sovereign AI. What we got wrong about a feature. How we think about safety. The tone is more reflective than promotional.

When we ship something interesting, the engineer who built it writes a few hundred words about why we chose this approach over the obvious one. These are honest about trade-offs; they are not press releases.

Customers and contributors writing what they learned.

A short thought, sometimes opinionated, often unfinished.

Bigger ideas; sovereignty, ethics, where AI is heading.

Why we made a decision, what surprised us, what failed.

Most company blogs are content marketing. Generic articles, written by interns or by AI, designed to rank in search. This one is different. The posts are written by the people building the technology, in their own voice, on the topics they actually find interesting. Some posts are short. Some are long. Some are a bit too technical for everyone. We do not chase rankings; we write things we would want to read.

Early findings and the dead ends that taught us something.

What broke, and the change that fixed it.

A clear note when something new arrives.

We write about the things we are actually working on. When a new feature is ready, we explain what it does and why we built it that way. When something breaks, we explain what happened and how we fixed it. When we discover something interesting in our research, we share it so you do not have to rediscover it yourself.

Long reads and short notes from the people building Dweve. Pick one and start reading: every post is the full text, with no sign-up and no tracking.

No marketing gloss. We write when something ships, when something breaks, and when a research thread is worth your attention. If it is quiet here, it means the team has its head down shipping.

Use it as a source library: AI safety, sovereignty, hardware, privacy, infrastructure, and practical AI explanations.

Use it for context, send a topic when a gap matters, or keep reading from the current list. Every post carries its author and date.

A complete archive of essays, explainers, engineering notes, and research arguments from the original site.

No gated posts, no tracking scripts, no fake publishing cadence. Every post in the archive opens in full, with the author and the date on it.

The full original blog set is back here: long essays, explainers, engineering notes, and research arguments.

No tracking scripts, no gated posts, no fake cadence. Pick an article and read it.

The technical view keeps implementation and architecture posts near the surface, without hiding policy and safety context.

Engineering and architecture posts stay close to the top, while the rest of the archive remains one click away.

Architecture, policy, safety, and practice.