Back-Office Automation × Context Graph
The Connection
Back-office automation is outcome-shaped work across HR, IT, finance, and operations. The hard part is rarely the LLM — it is knowing who someone is, what they already have, what policy allows, and what actually changed in each system. The context graph is the shared memory that makes those questions answerable without a human opening six tabs. ^[inferred]
Where They Co-occur
Agent-first service desk pages, MCP workflow generation, and market landscape notes all assume that automation depth is bounded by context depth. The wiki links back-office ambition to graph primitives whenever it discusses Init Intelligence expanding beyond IT.
Cross-cutting Insight
If Init Intelligence wins IT as the wedge, the expansion path is not “more generic workflows” but a richer graph that follows the employee lifecycle across tools. Back-office automation without graph investment becomes brittle RPA; with graph investment, the same runtime can absorb adjacent functions because the entities (person, device, app, contract, ticket) stay stable even as workflows diversify. ^[inferred]
Tensions and Trade-offs
- Scope creep: buyers will ask the graph to solve data quality they never fixed in the CMDB.
- Privacy: richer graphs increase blast radius if access control is wrong.
- Build order: early teams may fake a graph with integrations only; at some scale, explicit modeling wins. ^[ambiguous]
Open Questions
- Which non-IT domains share enough entity overlap (identity, assets, vendors) to reuse the IT graph without a greenfield project?
- When should Init Intelligence recommend professional services vs productized connectors for graph completeness?