← What we do

Amplify · phase 4

Managed Skills

Your method, voice and rules, versioned and delivered into every person’s AI. The human stays on the judgement.

The problem

A lot of valuable work should stay human-led: a tender response, a client proposal, a judgement call. But it still needs to be done your way every time, not reinvented by each person from scratch. Left to personal AI chats, that method is inconsistent, unversioned, and walks out the door when people do.

What you get

  • Your best method (the voice, the rules, the structure) packaged into governed skills, published to your organisation’s Claude library
  • Central method, local control: the skill hands each person a clean, governed output, and they tailor it with their judgement
  • Your existing ad-hoc AI habits ported into versioned, governed skills
  • A curated library that compounds in value instead of going stale

How it’s governed

Skills are versioned and centrally curated: one source of truth for “how we do this here,” with the human always kept on the judgement.

This is the human-led half of the two-tier build that sits alongside your Managed Agents.

Why it’s credible.It’s the same discipline Orange AI runs on. A governed, versioned library of 100+ skills is how our own work stays consistent as it scales, and every one of them carries a formal draft-to-active gate before anyone relies on it.

Riverlight Pty Ltd · Skill librarySample data
Tender response, our way v2.3
Structures every tender to the house method: win themes, compliance matrix, your tone.
Sales · 14 usersActive
Client proposal builder v1.8
Drafts proposals in your voice with your commercial rules baked in. Human signs off.
All staff · 31 usersActive
Site-report writer v1.1
Turns field notes into your standard report format, terminology and all.
Ops · 9 usersActive
v1.0
Draft
v0.9
Draft
Your library grows
with every engagement

Versioned, statused, and curated. A skill only goes active once it’s been reviewed against your method.