Learning-Budget-Optimizer

Budget Intake Agent receives three inputs: total budget, the team roster with roles, and the upcoming project pipeline (which certs are needed when). It normalises this into a structured payload that all downstream agents consume. Work IQ contributes project deadline signals here — which projects ship next quarter, and which certs they require. ROI Scoring Agent runs the Fabric IQ formula over every employee-cert pair. It also pulls from Foundry IQ to fetch historical pass rate data from the synthetic team learning report. Output: a ranked list with ROI scores, skill gap sizes, and project alignment scores. Allocation Agent is a greedy knapsack solver — iterate from highest ROI to lowest, allocate budget until it runs out. It also flags "near miss" employees who just missed the cut and could be funded with a small budget increase. This is the agent that generates the "if you add $2,400 more, you cover EMP-004 who is on your highest-priority Azure migration" insight. Plan Generator Agent takes each funded employee and generates a personalised study schedule, using Work IQ to fit study blocks around their calendar load. This reuses the Study Plan Generator pattern from the baseline challenge spec. CFO Insights Agent is the narrative layer — it translates the allocation output into a business story: "We invested $9,800 in 4 certifications this quarter. Projected pass rate is 76%. All funded employees are on high-priority delivery tracks. Expected revenue-impacting projects unblocked: 2." This is what leadership actually reads.

The three IQ layers, used precisely Fabric IQ holds the ROI scoring ontology — the relationships between employee, role, certification, skill gap, pass rate, and project demand. It's the semantic layer that makes the scoring formula portable across all five agents rather than duplicated in each. Foundry IQ indexes three synthetic documents: the Engineering Certification Guide (which certs map to which roles), the Quarterly Learning Performance Summary (historical pass rates by cert and role), and a Project Pipeline doc (which upcoming projects need which skills). Every allocation decision cites the source. Work IQ contributes project deadline signals to the Budget Intake Agent, and calendar/focus-time signals to the Plan Generator. It's what makes the system aware that "Q3" isn't just a label — it's a specific window with real capacity constraints.

The killer demo moment Open the simulator above. Set budget to $8,000. Show how only 3 employees get funded, ranked strictly by ROI score. Then drag to $12,000 — a fourth employee appears. Point to EMP-004 (Cloud Architect, AZ-305, ROI 74) — "this person was deferred at $8K but they're on the Azure migration that ships in October. At $12K they're in." Then hit "Generate CFO memo" — the AI writes the budget justification letter automatically. Judges who have ever sat in a budget meeting will immediately feel the value.

Built With

• agents
• aiml
• python
• react
• typescript
• uipath
• vectordb
• vercel