GrailHawk
A collectibles reference site used to test the ordinary standalone Site loop. grailhawk.com
- Live receipts
- 3
- Last observed
- Aug 14, 2026
- Latest recorded article
- How to value a collectible when there are no recent sales
Experiment overview
What this project is testing
The question, ownership, and evidence boundary in one place.
Current hypothesis
Evidence-led reference pages may become useful source material in later dated answer observations.
This is a prediction to test, not an outcome claim.
Ownership disclosure
We built these sites to test AgentReady on ourselves. Everything we do to grow them is public.
Owned and operated by the AgentReady team as a ratified Labs dataset. This is not presented as independent customer proof.
Daily run receipts are missing
This page can prove the publications and product events below. It cannot prove that a planned topic shipped, or explain a no-output run, until the daily operator ledger is added to the public read model.
Published work
What is actually live
Every approved same-domain canonical in the current public feed, newest first.
Publication ledger
3 live canonicals recorded
Newest receipt Jul 29, 2026
These are the articles this public source can prove are live. If a canonical is not in this list, this page does not claim it shipped.
- How to value a collectible when there are no recent sales
https://grailhawk.com/blog/value-a-collectible-with-no-recent-sales
Page measurement pending - How to compare collectible prices without mixing bad comps
https://grailhawk.com/blog/compare-collectible-prices-without-bad-comps
Page measurement pending - Publish a "Which Rolex Cosmograph Daytona would you recommend?" comparison vs WatchBox
https://grailhawk.com/watches/rolex-cosmograph-daytona
Page measurement pending
Counts exact live canonicals from AgentReady verification and Site publication receipts. Operator editorials remain labelled in the source ledger and are not treated as recommendation fixes. Per-post GSC and top queries use the connected 28-day Search Console window.
Measurements
What the sources observed
Dated product and Search Console observations. Movement is measured since publication, never claimed as caused by it.
Days in program
28
since July 17, 2026
Recommended fixes
91
current queue
Verified fixes
1
1 marked applied
AI-answer presence
0 of 292 answers
- google0 of 127
- openai0 of 113
- perplexity0 of 52
Assistants answer differently run to run. These are counts across dated runs, not a rate we promise.
Readiness grade
Not measured on this source yet.
Search and publication ledger
GrailHawk in Google Search Console
Dated first-party measurements from the connected AgentReady account. Results are measured since publication where a canonical has page-level data, never because of it.
- Tracked posts
- 3
- AgentReady records
- Impressions
- 1.6K
- 37 dated rows
- Clicks
- 0
- synced Aug 14, 2026
Evidence activity
One timeline, clear event types
Publication receipts and product evidence are combined by date but remain separately labelled. Newest evidence appears first.
1
Published
A live canonical receipt exists.
2
Applied
A recommendation was marked applied.
3
Verified
A later recrawl found it live.
4
Measured
A later dated source observation exists.
Measured
perplexity answer check
0 of 12 answers mentioned the brand in this dated run.
Measured
google answer check
0 of 24 answers mentioned the brand in this dated run.
Measured
openai answer check
0 of 18 answers mentioned the brand in this dated run.
Measured
perplexity answer check
0 of 12 answers mentioned the brand in this dated run.
Measured
google answer check
0 of 24 answers mentioned the brand in this dated run.
Measured
openai answer check
0 of 15 answers mentioned the brand in this dated run.
Published
How to value a collectible when there are no recent salesA same-domain live canonical was added to the public publication ledger.
Published
How to compare collectible prices without mixing bad compsA same-domain live canonical was added to the public publication ledger.
Measured
perplexity answer check
0 of 17 answers mentioned the brand in this dated run.
Measured
google answer check
0 of 34 answers mentioned the brand in this dated run.
Measured
openai answer check
0 of 35 answers mentioned the brand in this dated run.
Measured
perplexity answer check
0 of 11 answers mentioned the brand in this dated run.
Show 10 older evidence events
Measured
google answer check
0 of 22 answers mentioned the brand in this dated run.
Measured
openai answer check
0 of 22 answers mentioned the brand in this dated run.
Published
Publish a "Which Rolex Cosmograph Daytona would you recommend?" comparison vs WatchBoxA same-domain live canonical was added to the public publication ledger.
Verified
Publish a "Which Rolex Cosmograph Daytona would you recommend?" comparison vs WatchBox
A later recrawl confirmed the artifact was live.
Recrawl failed
Publish a "Which Rolex Cosmograph Daytona would you recommend?" comparison vs WatchBox
The recrawl did not find the artifact live at this time.
Applied
Publish a "Which Rolex Cosmograph Daytona would you recommend?" comparison vs WatchBox
Recommendation marked applied with content-item.
Measured
google answer check
0 of 11 answers mentioned the brand in this dated run.
Measured
openai answer check
0 of 11 answers mentioned the brand in this dated run.
Measured
google answer check
0 of 12 answers mentioned the brand in this dated run.
Measured
openai answer check
0 of 12 answers mentioned the brand in this dated run.
Verified changes
What changed and what happened later
Recommendation records stay separate from editorial publication receipts. Each record shows its recrawl ladder and later observations.
comparison article
Publish a "Which Rolex Cosmograph Daytona would you recommend?" comparison vs WatchBox
Since this fix went live, the brand was mentioned in 0 of 24 answers (before: 0 of 2).
Measured since then, not necessarily because of it.
Method and limits
How to read this evidence
Known limitations and system gaps stay visible beside the results.
Limitations
- A citation, a brand mention, and a recommendation are different observations.
- This is one owned niche site, not evidence of typical customer outcomes.
Open system gaps
- Durable autonomous cycle records are not in the public read model yet.

