Software for complex work. Built around people.
Grid360 brings software engineering, artificial intelligence, automation and interface design together. We build connected systems around real operational problems. NewsMind, our first product, applies that approach to the newsroom.
- 01
- Platform shipping
- NewsMind, in newsrooms now
- 03
- Foundations
- Archive, brand, permissions
- 04
- Opinions we hold
- Each one costs us something
A foundation first, and a platform on top of it.
Three things sit underneath everything we make: an archive that describes itself as material arrives rather than when somebody needs it, a brand that resolves at the moment of use rather than being applied by hand, and one model of who may do what that every surface obeys. None of those three are specific to news. They are what a media organisation needs before it can automate anything safely.
NewsMind is that foundation turned into a newsroom: nine studios covering the wire, the archive, broadcast production, live clipping, voice and brand, running as one installation inside your building. It is the first product on the foundation. It will not be the only one, and the next is being built against the same three things rather than beside them.
- Shipping NewsMind
- In build The second platform
- Open What comes next
- An archive that describes itselfMaterial is described as it arrives, not when somebody needs it.01
- A brand that resolves liveApplied at the moment of use, never remembered by hand.02
- One model of who may do whatEvery interface obeys the same permissions.03
None of the three is specific to news.
Four opinions, held expensively.
Automation should show its reasoning.
A system that produces an answer without showing how it got there is asking a newsroom to take responsibility for something it cannot inspect. That is not a reasonable thing to ask of an editor whose name goes on the output. So every automated judgement in our software arrives with what drove it, what else was considered, and a way to overrule it in one action. This is more work to build and it makes the product look less magical in a demonstration. It is the only version that survives contact with a working desk.
A tool for journalists has to be able to say “I don't know.”
The easiest thing to build is a system that always returns something. It demonstrates well, it never disappoints in a meeting, and it is quietly useless — because a result that might be invented and a result that is verified look identical. We spent a great deal of effort on the opposite: on the empty slot, the refused match, the article that will not be written, the answer that says plainly it cannot be given. Those are the parts that make the rest of it trustworthy.
Arabic is a first language, not a localisation task.
Most broadcast software is built in one language and adapted afterwards, and the adaptation always shows: letters that do not join, lines that reverse when a number is added, captions sized for a script that reads differently. We built the other way round. Scripts are written in Modern Standard Arabic, text sets correctly in every graphic and subtitle, and transcription was designed around the specific failure where machines produce fluent Arabic nobody said. None of that is a feature list item. It is the difference between software an Arabic newsroom can use and software it works around.
Automation is something you switch on, never off.
Our software can run a route unattended — publish a clip, write a story above a newsworthiness threshold, send a package on. What it will not do is arrive that way. Every automatic route starts off, and turning one on is a deliberate act by somebody who holds that permission, recorded when it happens. The reason is not squeamishness about automation; it is that a desk in its first week has no basis for the decision and a desk in its third month does. Shipping it the other way round would mean the most consequential setting in the product is the one nobody chose.
Start with the part that hurts most.
Nobody adopts nine studios in a month, and we would not recommend trying. Most newsrooms start with one problem — the archive nobody can search, the live clipping that takes six minutes a clip, the brand that drifts — and add the rest once the first one is working. Tell us what your desk actually struggles with and we will tell you honestly whether we help.
Let's build what comes next.
Whether you are modernising a workflow you already have, introducing automation for the first time, or building something new, tell us what the work actually looks like — and we will say honestly whether we are the right people for it.