# Attention Game - Work Last updated: 1 October 2026 Six lines of work, and one method underneath all of them: embed in an operation, work out what it actually does, then build the thing that does it repeatably. The domains differ; the method does not. Client engagements are described as patterns rather than named, because the detail belongs to the clients. ## Where the work starts Attention Game is a content and creative strategy practice equipped with AI and agentic tools. The domain expertise underneath it is specific: two decades building systems that produce content for platforms - YouTube, TikTok, Facebook, Instagram, and the ones that came before them - and understanding what each one actually rewards. That is the lens on everything here. A content system that does not understand a platform produces volume. One that does produces performance. So strategy comes first - what the audience is doing, what the business is for, what should be made - and the systems are built to serve it. It is also the difference between this and an AI consultancy. An AI consultancy starts with the technology and looks for a use case. Here the starting point is the operation and the platforms it publishes to, and the tools are chosen afterwards to fit. ## Content intelligence **What it is.** Performance monitoring tuned to a specific domain, with outlier detection and trending topic alerts, covering all major platforms and paid ad libraries. **How it works.** Every piece of content is scored against its own channel's normal rather than a cross-industry benchmark. A video that does 40,000 views on a channel that usually does 8,000 is the signal; the same number on a channel that usually does 400,000 is a problem. Industry benchmarks cannot see either. **What it produces.** - Outlier detection, up and down, against the channel's own baseline - Which formats, lengths, hooks and subjects are carrying performance, and by how much - Trending topic alerts inside the domain, rather than general trend feeds - Coverage across major platforms and paid ad libraries, so organic and paid activity read on the same surface **Who it is for.** Teams publishing enough that the pattern is no longer visible by eye, and who need production decisions to follow evidence. ## Dashboards and reporting **What it is.** Multiple data sources across social, marketing and sales aggregated into one hub, with views partitioned per stakeholder and notifications on a set cadence or on triggered metrics. **How it works.** The numbers that describe a business are scattered across the platforms, the marketing stack and the sales stack, and nobody sees the whole picture without assembling it by hand. Everything is brought together into one hub, and then partitioned: a board view, a commercial view, a partner view, each reskinned so it reads as its own thing rather than a filtered version of someone else's dashboard. **What it produces.** One source of numbers instead of several, and reporting that arrives on its own. Notifications run on a set cadence, or fire when a specific metric crosses a threshold, so the people who need to act find out without having to go and look. ## Sponsorship measurement **What it is.** AI agents analyse video, audio and social posts to record where a brand appears across a rights holder's owned, partner and athlete channels. Built with [Implicitly](https://www.implicitly.io). **How it works.** A sponsor's value is not delivered in one place. It accumulates across the rights holder's owned channels, its partners' channels and its athletes' channels, in video, in audio, in copy, and in the moment each appearance sits inside. The system reads all of it and records every form of appearance - the logo, and the copy, social copy, name checks, endorsements, implied endorsements and products around it. **What it produces.** Evidence a commercial team can price and renew against, and a record a partner can see for themselves. Our thinking on this is in [point-of-view.md](https://attngame.com/point-of-view.md). ## Knowledge bases **What it is.** Institutional knowledge systems that give a business one source of truth, built so that both colleagues and AI agents can use them. **Three patterns we build.** - **Product knowledge** - e-commerce ranges structured so that specifications, claims, ingredients and positioning are consistent everywhere they appear. - **Canonical source of truth for regulated industries** - where the same fact has to be stated the same way every time, and a wrong version is a liability rather than an inconsistency. - **Departmental and company knowledge bases** - covering everything from onboarding new staff to troubleshooting day-to-day content questions. **Why it matters now.** Knowledge that lives in people's heads was always a risk. It is now also the thing standing between a company and useful AI, because an agent can only be as good as what it can read. ## Personal brand engines **What it is.** Tools for creators that scale content volume and keep publishing consistent, built on voice notes, trending topic alerts and a codified tone of voice. **How it works.** The bottleneck for most people with something worth saying is not ideas, it is the distance between having a thought and publishing it. The engine shortens that distance: thinking is captured as voice notes and interviews, a codified tone of voice holds the output in the person's own register rather than flattening it into generic copy, and topic alerts tell them when a subject is live. **What it produces.** A sustained publishing operation that runs at the volume the platforms reward, without the person having to write everything or sound like someone else. ## AI transformation **What it is.** We audit the operation, upskill the team in AI tools and techniques, and make processes and data legible to AI. Then we co-create bespoke tools with the team, to solve problems, improve processes and develop new products and services. **How it works.** The audit maps what the team actually does, where time goes and where knowledge is trapped. Upskilling runs alongside it, so the people doing the work can see what the tools are now capable of. Codification turns the processes that survive the audit into written, structured form. Then the building is done with the team rather than for it - the tools that remove the specific work the audit found, not a generic platform. **What it produces.** A company whose knowledge and processes are legible to both people and agents, a team that can use the tools without being led by the hand, and a small number of tools doing real work rather than a pilot.