IBC 2026 · RAI Amsterdam · 11–14 September 2026
We work inside your pipeline. SRT contribution, ffmpeg and GStreamer transcode, HLS and DASH packaging in GPAC, MoQ delivery, and the storage and compute underneath all of it.
Four things teams call us in for.
Two to five of our engineers on your backlog: the features your team has not had bandwidth to start, built inside your codebase and your review process. And when the next bet needs de-risking first, a proof of concept in weeks: can MoQ get you under a second, does watermarking survive your ladder.
Your video product, kept shipping after the original team moved on: the ffmpeg and GStreamer upgrades nobody wants to own, codec, dependency and security bumps, the regression suite that catches a broken encoding ladder before a customer does, and a backlog that actually shrinks.
The glue work nobody budgets for: your video stack wired into the CMS, the DRM licence server, the EPG feed and the CDN, down to the kernel driver and the embedded box when that is where the problem lives, and the on-call while it beds in.
S3-compatible object storage, transcode capacity that scales with the queue instead of sitting idle between events, and deployments on Scaleway or your own metal for teams whose media has to stay in the EU.
IBC Accelerator 2026
Q-Stream: Trusted Live Media Infrastructure
The IBC Accelerator programme puts vendors and broadcasters together to build something real against a show deadline, in public, with the results demonstrated on the floor. We are one of the engineering participants.
Q-Stream is about video that degrades gracefully instead of failing: adapting the stream from predictive network insight, compressing semantically rather than uniformly, and prioritising the parts of the picture that carry the meaning, so a feed stays usable at bitrates where it would normally fall over.
Our part is delivery and deployment: the SRT contribution leg into the Norsk media engine, MoQ playback with a self-hosted relay so nothing depends on the venue internet, per-segment C2PA verification in the browser player, and the one Helm chart that runs the stack on cloud, on-prem Talos, or a fully air-gapped rig.
With the Associated Press, BFBS and the IET as Champions, and Tesla Technologies, Norsk and InReality as the technology partners alongside us.
Who we work with today, and what the work is.
Cloud platform services
IPTV / digital signage platform
Web conferencing platform
Prototypes
Video and cloud engineering, as their team rather than as a vendor at arm’s length.
Two shapes. The operational answers are below them, because that is what actually stalls a deal.
Two to five of our engineers inside your team: your sprints, your repo, your standups. You direct them the way you direct your own people.
Fixed scope, named outcome. Worth it when the problem is well understood and you need it built without adding permanent headcount. If you hold the client relationship, we deliver under your name.
We build with coding agents every day. None of your code feeds someone else’s model.
Every agent runs in a disposable sandbox: an isolated checkout, scoped credentials, no route to production. What it produces arrives as an ordinary merge request, and the engineer who owns the change reviews it like any other.
The models run on our GPUs, or on yours. Open-weight LLMs served next to the repo, so prompts, diffs and embeddings never cross to a third-party API and there is nothing new for your DPA to cover.
The gain is real but not uniform. Prototypes, test suites, migrations and documentation come back in days instead of sprints; the judgment work still takes what it takes. The visible difference is that the work teams used to skip now gets done.
Small team, all in one office. You talk to the people writing the code, and the person who scopes the work is the person who reviews it.
Thirty minutes, no deck. Tell us what the pipeline is doing and we will tell you whether we have fixed that before.
