Devectordevector
Book a call

Work

Clients, and the shapes of work we run.

We're light on specifics here — most of our work sits inside enterprise teams under NDA. What we can show is who we work with, the themes that thread through our engagements, and the kind of project we typically take on.

Clients

Working with enterprise teams across media, content and hospitality.

Most engagements are multi-quarter and ongoing. The list below is who we've shipped or are shipping for — happy to make introductions where appropriate.

Themes

Three threads that run through our work.

The exact systems differ by client. The shape of how we engage looks similar — own infrastructure, own data, careful rollout, real adoption.

Theme · 01

Private AI for whole companies

A shared, secure AI workspace for everyone in the business — running on infrastructure you control, with the models, integrations and agents that suit your teams.

Theme · 02

AI woven into operations

Agents and copilots inside the systems your team already uses — drafting comms, triaging requests, summarising the things nobody has time to read.

Theme · 03

AI inside the product you ship

Search, vision, generation and summarisation, embedded inside the platforms you sell to your own customers — production-grade, observable, owned by you.

Engagement examples

What a typical engagement looks like.

Anonymised illustrations, drawn from real engagements. Different industries, similar shape. If yours looks like one of these, we should talk.

Engagement 01
Live

A private AI workspace for a national media business.

A self-hosted, multi-model AI platform stood up for a large content company — wired into their identity, their internal sources, and their editorial and marketing workflows. Phased rollout, hands-on training, retainer to keep it growing.

  • Private workspace inside the client's own cloud
  • Multiple models available, switchable mid-conversation
  • Workspace-level governance — SSO, RBAC, audit
  • Phased rollout across teams with adoption support
Workspace · multi-team rollout
Engagement 02
Live

AI inside the newsroom — built for the pace of news.

A private workspace and a set of editorial agents for a wire-service business — fast turnarounds, internal context plumbed in, and source material that never leaves the perimeter. Designed to fit the way newsrooms actually move.

  • Editorial-grade workspace with newsroom-specific agents
  • Internal style, archive, and briefings as context
  • Per-team usage and adoption visibility
  • Hands-on enablement for journalists and editors
Workspace · editorial agents
Engagement 03
Rolling out

AI for every department — across countries and languages.

A private AI workspace rolling out across an international hotel group, department by department — guest comms, marketing, operations, finance — with multi-language support baked in and clear governance for a multi-country business.

  • Multi-property, multi-language workspace setup
  • Agents shaped to operations and guest comms
  • Phased department rollout with training
  • Governance shaped for cross-border operations
Workspace · multi-language
Engagement 04
In progress

AI woven into a content platform serving customers worldwide.

An AI layer wrapped around a large content platform — search, metadata, contributor support and editorial review — designed to stand up next to a content library at internet scale, and a private workspace for the teams who run it.

  • Vision-model pipelines tuned to a content taxonomy
  • Internal workspace for support and editorial
  • Architecture sized for hundreds of millions of assets
  • Production-grade with evals and observability
AI inside product · vision pipelines

We're picky about who we take on. We do better work when the engagement is with leaders who know what they want from AI — and are honest about what they don't. If that's you, drop us a line.

Next step

Sound like the kind of engagement you're after?

Most engagements start with a 30-minute call. We'll talk through what your team needs from AI, walk you through how we typically run this kind of work, and tell you honestly whether we're the right shop for the job.

Book a callor email dev@devector.io