Bluebird Horizon: APS workforce planning

APS-wide headcount by job family from FY2018 to FY2025, AI task exposure by job family, and headcount by classification.

FY2025 endpoints are the APSC's published job family shares. Earlier years and the classification view are modelled from published distributions.

All 102 APS agencies, each with its published headcount at 30 June 2025, a headcount outlook, its job family mix and its sharpest gaps.

Everything beyond headcount is modelled for the demonstrator. Tier 1 data makes this view possible before any agency system is connected.

What the AI transition is worth to each agency by FY2030 once transition costs are paid, shown as capacity that can be redeployed to priority work. Figures are full-time-equivalent capacity, not positions.

The model's coefficients are calibrated to cited evidence and are editable assumptions in the platform. See Data and method.

A curated skills baseline mapped to the APS Job Family Framework, with coverage by job family, the fastest-moving skills and a check of whether the external market can fill each gap.

Skills are extracted from position descriptions with a confidence score on each match. Matches below the threshold queue for human review. Extraction from job descriptions, role profiles and learning records comes with the pilot and build phases.

Demand and supply projected to FY2030 from the published 30 June 2025 baseline, with 50% and 90% uncertainty bands, and the gap between them charted directly.

Models are validated point-in-time: scored only on data they could not have seen. A detected regime break is marked on the chart.

APS demand set against national supply, using Jobs and Skills Australia's Occupation Shortage List, Internet Vacancy Index and Employment Projections, education completions, skilled migration and the ACS Digital Pulse.

Published figures are tagged as published. Everything derived is labelled as modelled, and the weighting behind the contestedness score is stated.

• The largest projected skill gaps, ranked by time to impact.
• Single-point dependencies: critical capabilities held by fewer than five people in an agency.
• Options to close each gap across four levers: buy (recruit), build (reskill), borrow (mobility and surge) and bot (automate task content).
• A generated workforce risk register. Planners assign the owners and treatments.

How Singapore, the United Kingdom, the United States and OECD guidance compare with Australia, each with the lesson it carries for the APS.

Adjacent-skills matching connects people in transforming roles to future-critical roles they are already most of the way to. A pathway is recommended only when its skills-to-train list is short enough to schedule.

Every recommendation is reviewed by a person before it reaches a conversation with an employee. Aggregated skills-to-train lists can also become a commissioning signal for micro-credentials and graduate streams with university partners.

Sliders for AI adoption speed, budget growth, attrition and reskilling investment, with outcomes recomputed as the levers move.

Every saved scenario is stamped with its assumptions and model version, so every run is reproducible and auditable.

A drafted workforce strategy for each of the 102 agencies, grounded on that agency's own scorecard. A person owns the final document.

In the demonstrator the drafts are pre-generated. In production they are drafted live through the model gateway on the same grounding.

A whole-of-government AI transition scorecard and a one-page scorecard for each agency, computed from the model, with every figure traceable to its page. Worked examples include a strategic workforce plan extract, a quarterly workforce risk brief and a surge readiness assessment. Everything prints for the executive pack.

A chat interface that draws on the APS dataset, the daily news feed and your own library. Ask for a report, such as "write this up as a Word report", and you get a formatted document with charts, saved to the knowledge base.

Chat and report generation are part of a free 7-day trial for Australian government agencies.

A daily feed of news and Australian parliamentary material, including Hansard, Senate estimates and committee hearings where agencies answer for AI adoption and the workforce. The original source is linked on every item.