BlockSentient publishes weekly observations of recognized AI-agent requests to review pages across the six Veracity Media Network directories.
These reports measure activity on pages operated by our network. They do not measure traffic to vendor websites, and they do not establish whether an AI system used, cited, recommended, or relied on the requested content.
Our web servers record ordinary request information when pages are accessed. A daily process compares declared user-agent strings with a maintained list of recognized AI-related crawler and retrieval patterns.
For each tool in the directory, we publish a weekly report at /ai-report/{tool-slug}/.
Each report shows:
Reports cover the most recent complete Monday–Sunday UTC period and are regenerated every Monday at 02:00 UTC. A report may contain no observed activity for a particular week.
We classify requests using the declared user-agent string and a maintained manifest of documented AI-agent patterns.
Live retrieval identifies recognized user-agent patterns associated with on-demand retrieval or search activity.
Training/index identifies recognized patterns associated with scheduled crawling, indexing, dataset collection, or model-related collection.
These labels describe the matched user-agent pattern. They are not independent verification of the requester’s identity, internal purpose, or subsequent use of the page.
User-agent strings can be changed or spoofed. AI systems may also retrieve content through unrecognized agents, shared infrastructure, browser-like user agents, cached copies, search indexes, or other sources that our process cannot identify.
A recognized request does not prove:
The reports are server-log observations, not citation tracking or prompt-response testing.
Standard server logs may include an IP address, timestamp, requested host and path, response status, referrer, and browser or user-agent information. Raw server logs are retained for up to 90 days for security, reliability, and measurement purposes.
For AI-activity analysis, we create minimized event records containing only the fields needed to classify and aggregate recognized requests. These derived records exclude IP addresses, query strings, referrers, and full user-agent strings and are retained for up to 12 months.
Public weekly reports and research datasets contain aggregate counts rather than raw request records. Aggregate reports and published research may be retained as part of the network’s historical record.
Further details are available in the Privacy Policy.
BlockSentient measures requests to pages published on BlockSentient and the other directories operated by Emerging S.R.O. We are therefore reporting activity observed on infrastructure that we operate. Results should be interpreted within that scope and may not represent activity across the wider web.
The same collection, classification, and reporting method is applied regardless of whether a vendor has an affiliate relationship, has purchased a separate service from Emerging S.R.O., or has no commercial relationship with us.
Emerging S.R.O. may offer separate paid research or documentation-audit services. Payment cannot alter recorded request counts, AI Visibility results, CTI scores, rankings, factual corrections, editorial coverage, or eligibility for a Vendor Verified badge.
If a directory review contains a factual error or omits a relevant public source, please send us the correction and supporting vendor-owned documentation. Corrections are evaluated under the evidence rules described in our Methodology.
Vendor Verified means that a vendor has confirmed the factual representation of its listing. It is not a product endorsement, security certification, compliance certification, or guarantee of AI visibility.
Researchers and journalists may reference our public aggregate reports and published datasets subject to the stated methodology, limitations, license, and attribution requirements.
BlockSentient is operated by Emerging S.R.O., a two-person company based in Brno, Czech Republic.