An approach to the five accountabilities, mapped against a library system I have already built and handed over once.
Media Assets & AI Automation Coordinator · October 2024 to June 2026 · iconik
assets organised and tagged by hand before any automation. That pass is what produced the schema
Classification-first model, so each asset carries only the fields relevant to its type
AI tagging with a human in the loop: per-field accept, edit and bulk approve
Versioned precision and recall gate deciding when tagging writes back unattended
Plain-language asset search inside Slack, where the teams already worked
Runbooks and troubleshooting guides maintained by non-developers
At IBC in September 2026, Backlight announced AI Discovery across iconik: AI-suggested metadata for a human to review and approve, and search in plain everyday language instead of exact tags. That is the direction those sessions were pointing at.
The practical value is simple. Before anything gets built custom here, I can tell you what the platform is about to ship natively.
A library the team searches, instead of a person they ask.
Every asset found is one that does not get commissioned twice. Re-shoots are the most expensive symptom of a library nobody can search, and they are charged to production budgets rather than to the DAM.
Requirements drawn from observed behaviour, validated before anything is built.
Build once. Reworking a live system costs more than building it, and it spends the credibility you need for the next change.
Fewer manual touchpoints between a shoot wrapping and an asset being usable.
Editor hours returned to editing. Most rework traces back to an unclear brief or unversioned feedback, and both are fixable with structure rather than effort.
A documented tool inventory and a costed consolidation case.
The one accountability with a direct dollar line. Seat recovery and tool overlap are realisable inside a twelve-month contract, which is what makes the rest of the roadmap easier to fund.
Documentation that measurably reduces support requests, and a retention standard that is applied rather than filed.
A team that stops queueing for one person. Retention tied to expiry also turns a compliance risk into a scheduled task.
Your Responsible Marketing & PR Policy, published July 2026, added dedicated sections on AI and digital integrity, truthful communications and climate honesty.
The roadmap is already defined. This is how I would get to its first decision with evidence rather than assumptions.
Understand the existing roadmap and where it has landed. Sit with the people who request footage and record what they actually do.
Volume, duplication, orphan rate, rights-field coverage, and which assets actually get used. Numbers, not impressions.
Taxonomy v1 on a single high-demand range rather than the whole library, so there is proof before there is a rollout.
Time to find, before and after. Then the rollout plan with the evidence attached.
Documentation written during the work, not after it. Documentation produced at the end of a project is documentation nobody trusts, and on a fixed-term contract it is the deliverable that decides whether the rest survives.
Twelve months means every decision is designed for handover: a named owner, a written standard, and a review date.
The measure I would hold myself to is that it still works six months after the contract ends. I have handed a system over once already. That is the part most implementations get wrong, and it is the part I would plan for from week one.
Every KPI on these slides is a measure I would put in place, baselined in the first ninety days so improvement can be shown rather than claimed.