Results by date
Showing 2026-08-27 to 2026-09-10. Latest measurement per prompt/platform within this range.
67.2%
Branded identification rate
Latest observations
2026-09-10 to 2026-09-10
45 hits / 67 runs across 25 cells. Latest measurement per cell.
The same share over the five branded prompts (P01 to P05), where the answer only had to identify the company.
Method: 45 hits over 67 runs across 25 prompt-and-surface cells, each at its most recent measurement (2026-09-10 to 2026-09-10).
18.8%
AI visibility rate
Latest observations
2026-08-31 to 2026-09-10
45 hits / 239 runs across 108 cells. Latest measurement per cell.
Includes mentions and recommendations, not only linked citations.
The share of measured AI runs whose answer identified, recommended or referenced Agnostify. This includes unlinked mentions and is not a linked-citation rate.
Method: 45 hits over 239 runs across 108 prompt-and-surface cells, each at its most recent measurement (2026-08-31 to 2026-09-10).
Recommendations and recorded links
Latest observations, 2026-08-31 to 2026-09-10: 0 cells have Tier B (recommended) as their best recorded result; 28 recorded citation entries contain an agnostify.com link. 3 recorded hits in this range carry source labels without exact URLs and count as no linked entry. Reconstructed observations are excluded from these counts.
Recommendation outcomes for each individual run and complete linked-citation coverage are unknown. No recommendation rate or linked-citation rate is inferred from mention tiers or missing citation lists.
Where Agnostify appears
- 5 of 25Prompts where Agnostify appears
Only measured prompts enter the total. A prompt counts when the latest measurement of any of its surfaces, AI or Google organic, identified, recommended or referenced Agnostify (tier A, B or C). Google organic runs only the two goal prompts, so a prompt can count on our own search listing alone.
- 6 of 6Surfaces that identify Agnostify by name
Only surfaces with a measured branded prompt enter the total. A surface counts when the latest measurement of any branded prompt (P01 to P05) on it reached tier A: the answer named Agnostify and described it correctly, or linked agnostify.com. A recommendation or a passing mention (tier B or C) does not count.
The scoreboard
Each cell is the best tier across the runs of that prompt on that surface at its most recent measurement. Only prompts with a hit on some surface are listed; a blank cell is a surface where that prompt has no hit yet, or one the frozen set does not run it on; "not run" is a planned cell no measurement has reached yet. All measured AI platforms, including Grok, contribute to the headline rates. Each recorded run counts once; Grok has one run per measured prompt.
"Google organic, spelling forced" is the branded query re-run with Google's spelling correction suppressed, because Google reads the brand name as a misspelling of a common word. It is shown here because it was measured, and it is deliberately left out of the rates and of the count of prompts where Agnostify appears: it is not the query a person types. Where a prompt shows a hit in that column and a blank under "Google organic", the brand ranks only when the correction is prevented from firing.
| prompt | ChatGPT | Perplexity | Gemini | Claude | Google AI Mode | Google organic | Grok | Google organic, spelling forced |
|---|---|---|---|---|---|---|---|---|
| P01branded agnostify Counts from tier A; 3 runs per v1 surface. Grok uses one run per measured prompt.
| ChatGPTA | PerplexityA | no hit yet | no hit yet | no hit yet | GrokA | ||
| P02branded Who is Agnostify? Counts from tier A; 3 runs per v1 surface. Grok uses one run per measured prompt.
| ChatGPTA | PerplexityA | GeminiA | ClaudeA | Google AI ModeA | GrokA | ||
| P03branded What services does Agnostify offer and what do they cost? Counts from tier A; 3 runs per v1 surface. Grok uses one run per measured prompt.
| ChatGPTA | PerplexityA | no hit yet | ClaudeA | GrokA | |||
| P04branded Case Media Company LLC Counts from tier A; 3 runs per v1 surface. Grok uses one run per measured prompt.
| ChatGPTA | no hit yet | no hit yet | Google AI ModeA | ||||
| P05branded Is Agnostify a real company, and who runs it? Counts from tier A; 3 runs per v1 surface. Grok uses one run per measured prompt.
| ChatGPTA | PerplexityA | ClaudeA | GrokA | ||||
| P06goal-verbatim agnostify Counts from tier A; 3 runs per v1 surface. Grok uses one run per measured prompt.
| no hit yet | no hit yet | Google organic, spelling forcedA |
19 measured prompts have no hit on any surface yet. Every run, hit or miss, is in the run records.
Measurements
day 14206 runs completed
The 48-hour update, every planned cell measured under unmatched conditions: Perplexity returned current prices on the branded price prompt in all three runs, every commercial discovery run missed (128 on the frozen AI surfaces, plus five of the nine Grok answers), ChatGPT and Gemini still state a $5,000 minimum that the services page does not carry, Claude repeats absence-of-records claims, and Google's branded organic result was absent from a first page that auto-corrected the query to agnostic.
Run record: https://agnostify.com/evidence/2026-09-10-visibility-summary.
- ChatGPT: 53 runs, signed in, Temporary chat, Unpersonalized, web search selected, visible model Instant, underlying version and configured locale unknown; one interrupted collection after the Mac locked was replaced and is excluded
- Perplexity: 53 runs, signed-in Chrome profile with the app's own incognito mode confirmed, Search mode, model control reads Model with the Pro advanced-search preview; underlying model and configured locale unknown; the first anonymous attempt was blocked and is excluded; conditions not matched to 2026-09-04, so directions are not comparable
- Gemini: 18 runs, signed in, Temporary chat verified, visible model Flash, automatic grounding with no separate search toggle, configured locale unknown; this cycle assigned P01 to P09 only, so 18 runs were planned and completed; the tracker's 48 is the full frozen-cohort capacity, and the omitted P10 to P25 cells were out of scope, not failed or unfinished; one run has source labels with no recovered destination; partial: 18 of 48; short cells P10 0 of 2, P11 0 of 2, P12 0 of 2, P13 0 of 2, P14 0 of 2, P16 0 of 2, P17 0 of 2, P18 0 of 2, P19 0 of 2, P20 0 of 2, P21 0 of 2, P22 0 of 2, P23 0 of 2, P24 0 of 2, P25 0 of 2
- Claude: 46 runs, signed in, incognito chat, web search on, Research off, visible model Opus 5 High as on 2026-09-04, configured locale unknown
- Google AI Mode: 21 runs, signed out; one new browser Incognito window opened for the sitting and reused across runs with a fresh AI Mode thread per query, not a fresh window per run, a deviation from the 2026-09-04 condition; AI Mode label, underlying model and country unknown; this cycle assigned P01 to P09 only, so 21 runs were planned and completed; the tracker's 29 is the full frozen-cohort capacity, and the omitted P10 to P25 cells were out of scope, not failed or unfinished; Google exposes only opaque google.com/goto redirect links, recorded by their exact href with the destination unknown; partial: 21 of 29; short cells P12 0 of 2, P16 0 of 2, P17 0 of 2, P25 0 of 2
- Google organic: 6 runs, signed out; the same reused Incognito window with fresh navigation per run, not a fresh window per run, a deviation from earlier sittings; country unknown; Google auto-corrected the branded query to agnostic and offered Search instead for agnostify, not clicked, so the first page held no agnostify.com result; not a matched condition against 2026-09-04 or 2026-09-06
- Grok: 9 runs, signed in on grok.com, Private Chat verified, conversation-history personalization off, no linked X account, visible model Fast, configured locale unknown; one completed answer per prompt plus a separate scoring chat that is not a measurement; adjunct, outside the frozen v1 comparisons
Shipped since the previous measurement:
- V2 release live 2026-09-09: new navigation, Work notes, capability pages, Lab and Start pages, guided intake, the optional project builder with /plan result pages, mobile scroll hardening
- /receipts hits-only view live 2026-09-07 (night): the grid lists only prompts with a hit, with the reach counts under the tiles
- Button unification (T7) and Apple-grade button polish (T9) merged 2026-09-10
- Grok summary import (2026-09-07) folded into the public rates and reach
day 119 runs completed
Grok summary import: four branded Tier A answers and five commercial misses, the same per-prompt tiers as September 4. Signed-in Private Chat with memory personalization off, as reported. Full answers and most exact citation URLs were not supplied. The X audit is dated August 31 and is excluded.
Run record: https://agnostify.com/evidence/2026-09-07-grok-summary.
- Grok: 9 runs, 9 reported answers imported from a scoring summary, not independently verified transcripts; one per prompt
day 1017 runs completed
Claude's branded price prompt moved from three misses to Tier A with current prices in every run, though one run overstated the audit credit terms. ChatGPT's invented $5,000 minimum did not recur. Both buyer-discovery prompts still miss.
Run record: https://chatgpt.com/s/cx_6a9ea11c4b9881918d34211bb5302143.
- ChatGPT: 11 runs, scoped to P02, P03, P09 and P12; partial: 11 of 53; short cells P01 0 of 3, P04 0 of 3, P05 0 of 3, P08 0 of 3, P10 0 of 2, P11 0 of 2, P13 0 of 2, P14 0 of 2, P15 0 of 2, P16 0 of 2, P17 0 of 2, P18 0 of 2, P19 0 of 2, P20 0 of 2, P21 0 of 2, P22 0 of 2, P23 0 of 2, P24 0 of 2, P25 0 of 2
- Claude: 3 runs, scoped to P03; partial: 3 of 46; short cells P01 0 of 3, P02 0 of 3, P05 0 of 3, P08 0 of 3, P09 0 of 3, P10 0 of 2, P11 0 of 2, P12 0 of 2, P13 0 of 2, P14 0 of 2, P15 0 of 2, P18 0 of 2, P19 0 of 2, P20 0 of 2, P21 0 of 2, P22 0 of 2, P23 0 of 2, P24 0 of 2, P25 0 of 2
- Google organic: 0 runs, not measured in this sitting; Corrected 2026-09-14: the branded runs previously published under GORG were re-filed to GORGX because the spelling correction was suppressed; observations are unchanged, and the unmodified query was not measured
- Google organic, spelling forced: 3 runs, scoped to P06
Shipped since the previous measurement:
- No site change between the two sittings; a scoped re-measurement of the cells that moved on 2026-09-04
day 8244 runs completed
The first full-cohort sitting, with corrected Google coverage detailed below: branded prompts identify Agnostify on ChatGPT, Perplexity and Claude, Gemini on one prompt, and every commercial prompt still misses on every surface.
Run record: https://agnostify.com/evidence/2026-09-04-visibility-summary.
- ChatGPT: 53 runs
- Perplexity: 53 runs, signed in with Incognito mode by decision; day zero ran logged out
- Gemini: 48 runs, signed in, Temporary chat, Flash, search grounding on, by decision; day zero ran under mixed models and search settings; lane B resumed after the first filing and completed the surface under the same date
- Claude: 46 runs
- Google AI Mode: 29 runs, location not verified as Los Angeles for P01 to P16 (Google's footer showed Ohio); P17 and P25 ran with the explicit Los Angeles parameter
- Google organic: 4 runs, Google reported Cuyahoga County, Ohio on every run despite the Los Angeles location parameter, so positions are not a Los Angeles measurement; P07 run 3 succeeded on the eleventh attempt, after the first filing. Corrected 2026-09-14: two of the three branded runs forced the spelling with nfpr=1 and moved to GORGX, and the third was intercepted by the autocorrect and rescored a miss, so the unmodified branded query was measured once here, not three times; partial: 4 of 6; short cells P06 1 of 3
- Grok: 9 runs
- Google organic, spelling forced: 2 runs, the two branded runs logged with nfpr=1, re-filed here 2026-09-14; observations unchanged
Shipped since the previous measurement:
- Profile copy re-pasted on X, GitHub, TikTok, YouTube, Instagram and Crunchbase (2026-09-03)
- Clutch and The Manifest profiles live and ranking on the branded query
- LinkedIn company page live (2026-09-02)
- /receipts published (2026-09-03)
- Homepage description names social accounts (2026-09-03)
day 66 runs completed
Branded organic Google stayed a Tier A result, though rankings were volatile. The commercial head term remained a miss.
Run record: https://chatgpt.com/s/cx_6a98a3a77c40819180705006332e5955.
- ChatGPT: 0 runs, not measured in this scoped run
- Perplexity: 0 runs, not measured in this scoped run
- Gemini: 0 runs, not measured in this scoped run
- Claude: 0 runs, not measured in this scoped run
- Google AI Mode: 0 runs, not measured in this scoped run
- Google organic: 6 runs
- Grok: 0 runs, not measured in this scoped run
Shipped since the previous measurement:
- No site change since day 4; a scoped branded-Google check only
day 448 runs recorded
ChatGPT's first measurement is a branded win, and Google AI Mode P04 flipped from miss to Tier A. Everything commercial still misses everywhere.
Run record: https://chatgpt.com/s/cx_6a95b542fa888191b2bbdd01cb647498.
- ChatGPT: 16 runs, partial run; 7 rows excluded or lost and 30 not attempted
- Perplexity: 2 runs, signup cap
- Gemini: 0 runs, the run was scoped to P01 to P09: 0 of the 27 it planned; planned here is the full cohort
- Claude: 0 runs, not reached
- Google AI Mode: 21 runs, the run was scoped to P01 to P09: 21 of the 21 it planned; planned here is the full cohort
- Google organic: 0 runs, CAPTCHA and browser-control mismatch
- Grok: 9 runs, P04 was not stated, and P06 was neither listed as a hit nor covered by the commercial-miss statement
Shipped since the previous measurement:
- Crunchbase profile live
- Instagram profile added 2026-08-31
- Sitewide positioning change 2026-08-30 (dropped the startups-and-small-businesses wording)
Some counts in this measurement were reconstructed from best-of-runs logs (runs set to the planned count, hits to the minimum the tier implies).
day 090 runs completed
Branded organic Google is already won, while commercial prompts are all misses. Branded AI answers work when web search is on.
Run record: https://chatgpt.com/s/cx_6a900d08ec308191bef9ed89ad91b684.
- ChatGPT: 0 runs, needs sign-in plus Temporary Chat
- Perplexity: 7 runs, logged-out query cap
- Gemini: 48 runs, mixed login, model and web-search conditions
- Claude: 0 runs, incognito redirects to sign-in
- Google AI Mode: 29 runs
- Google organic: 6 runs
Shipped since the previous measurement:
- Phase 1 site deploy, the same day as the run
- X, GitHub, TikTok and YouTube handles claimed hours before the run
Corrections and methodology archive
All-time archive. Not filtered by the selected measurement dates.
Agnostify is an AI and social media studio operated by Case Media Company LLC. See current services and pricing. The archive below records incorrect model outputs, not facts about the business.
12 of 18 logged items are still open.
GeminiP03open
Incorrect model output: Without web search it described Agnostify as a cloud infrastructure consultancy with invented pricing models.
First seen 2026-08-27.
GeminiP03open
Incorrect model output: With web search it said prices are published during onboarding rather than publicly.
First seen 2026-08-27.
Google AI ModeP04fixed
Incorrect model output: It invented a Case Media Group, an invented co-founder and a Case Automation division without connecting the LLC to Agnostify.
First seen 2026-08-27.
Fixed: the 2026-09-04 measurement no longer carried the claim.
Google AI ModeP04fixed
Incorrect model output: It attributed an unsupported vertical under a legacy brand name and made unsupported product and tool-wing claims.
First seen 2026-08-31.
Fixed: the 2026-09-04 measurement no longer carried the claim.
GrokP05fixed
Incorrect model output: It claimed no corporate record or Crunchbase profile could be found and that ownership is undisclosed.
First seen 2026-08-31.
Fixed: the 2026-09-04 measurement no longer carried the claim.
ChatGPTP05fixed
Incorrect model output: It stated public-record details as fact even though they were model claims, not facts published by Agnostify.
First seen 2026-08-31.
Fixed: the 2026-09-04 measurement no longer carried the claim.
GeminiP02open
Incorrect model output: It described Agnostify as a B2B software-as-a-service and consulting firm for data-driven decisions, which it is not.
First seen 2026-09-04.
ChatGPTP03fixed
Incorrect model output: It stated a $5,000-plus minimum project size taken from a third-party directory as Agnostify's reliable price, a figure not on the services page.
First seen 2026-09-04.
Fixed: the 2026-09-06 measurement no longer carried the claim.
ClaudeP03open
Incorrect model output: It quoted retired package names and prices (a $68,000 flat rebuild, a $100,000 bundle, a $112,000 component total and a 25 to 30 percent founding discount) that the live services page no longer shows.
First seen 2026-09-04.
PerplexityP04open
Incorrect model output: It resolved Case Media Company LLC to an unrelated media-production business focused on visual marketing for properties.
First seen 2026-09-04.
GeminiP04open
Incorrect model output: It resolved Case Media Company LLC to an unrelated media and entertainment company, then to Case Media Design, then to legal filings in other states.
First seen 2026-09-04.
ClaudeP05open
Incorrect model output: It claimed there was no independent coverage, no directory listing and no funding record, despite the Crunchbase, Clutch and The Manifest profiles.
First seen 2026-09-04.
PerplexityP03fixed
Incorrect model output: It attached prices to retired program names (a Full-Stack Operations & Social Rebuild at $68,000 and a Rebuild + Launch at $100,000) that the services page no longer lists under those names.
First seen 2026-09-04.
Fixed: the 2026-09-10 measurement no longer carried the claim.
GeminiP03open
Incorrect model output: It stated a $5,000-plus minimum engagement and a blanket audit credit toward any build; neither is on the services page.
First seen 2026-09-04.
ClaudeP03open
Incorrect model output: One run said every paid step credits toward the next; the services page restricts that ladder to ad review, teardown, pilot and retainer, with a separate capped credit on the operations audit.
First seen 2026-09-06.
Google AI ModeP02open
Incorrect model output: It stated that engagements come with guaranteed return metrics agreed before spending begins; no such promise is on the services page, and the directory it drew on was not verified.
First seen 2026-09-10.
ChatGPTP01open
Incorrect model output: It attributed a $5,000 project starting figure to The Manifest directory; an unverified directory discrepancy against the published $1,500 custom-project minimum, not a confirmed invented company price.
First seen 2026-09-10.
ChatGPTP03open
Incorrect model output: It again gave a $5,000 directory minimum as the reliable price, and one run resolved the name to Agnost AI and quoted that company's $0, $49 and $499 plans as Agnostify's prices.
First seen 2026-09-10.