How it works · powered by the Research Engine
Measure. Map. Make.
We measure how AI recommends across your category, map each gap to a specific content assignment, then make the content that closes it, and we rerun the same benchmark every month to see what moved.
+480%
verified AI wins in 11 weeks
Soapbox Bulletin, 5 → 29 of 80 queries
58 vs 34
verified #1 finishes vs the nearest rival
most recommended brand in its benchmark
+300%
wins, May to June
wealth-management client, from a zero baseline
Step one: measure
The AI names your real competition, and the reasons it picks them.
We do not guess at your competitor list, and neither do you. We ask the platforms your buyers use who they recommend in your category, and they name the field themselves. Sometimes it is who you expect. Often it is not: brands you never considered rivals are winning your buyers' questions right now. That field becomes the benchmark, the same buyer-intent questions against the same competitors, put to all five platforms every month, every result verified against each company's own domain.
A five-vendor fan-out, not one model's opinion.
Every buying question runs across ChatGPT, Claude, Perplexity, Gemini, and Grok at once. Asking one model gives you that model's take. Asking five shows you where they agree, where they split, and which names keep surfacing no matter who you ask. That spread is the difference between an anecdote and a read.
A recommendability judge that scores the why.
Each recommendation is scored on what actually moves an AI's choice: how clearly you answer the question, the reasons it gives for picking a name, the angles it leans on, and the patterns shared by whoever wins. You get a structured read on why you place where you do, not a single vanity number.
Grounded measurement, kept honest.
We read what the models say when asked to recommend, not language we coached into them, and we verify every result against each company's own domain. So the picture reflects the market's view of you, not your house marketing, and the win goes to the right company instead of a same-named firm absorbing it.
Step two: map
The benchmark shows what evidence the platforms weigh. We map every gap to a specific assignment.
The read is only useful if it turns into work. Each gap the analysis exposes becomes a concrete assignment: a page, a post, a social thread, a video script, pointed at a question your buyers actually ask AI. The measurement comes first and the writing follows it, so nothing gets made that the read did not call for.
Inputs
Discovery prompts
Brand knowledge
Perplexity research
SEO + traffic data
Measure · 5 AI platforms at once
ChatGPT
Claude
Perplexity
Gemini
Grok
after 2 human checkpoints
mentions · reasons · citations
9 reports
Share of voice
Head-to-head
Citation intel
Reasoning themes
GEO ↔ SEO
Keyword gaps
Content gaps
Exec KPIs
Prompts
Map · Intelligence
Theme normalization
raw reasons → canonical
Correlation engine
10 axes · winner patterns
Bridge themes
search intent → AI language
Make · Content
Blog & articles
editorial, in your voice
Web pages
rewritten for AI answers
Chatbot
cited answers · brand voice
Step three: make
We write the evidence that was missing.
We publish the evidence the benchmark says is missing, on your own domain, in structure machines can parse and cite, with a human editing and approving every piece before it goes live. As that library grows, we rerun the benchmark to see whether the recommendation moved.
An editorial desk that writes the fix.
The read is half the job. Our editorial desk writes the pages, posts, social, and scripts aimed at the exact gaps the analysis found, on your own domain, in structure machines can parse and cite. Most tools stop at telling you what is wrong. We write what is missing.
Two human pauses, by design.
Nothing goes out on autopilot. The work stops twice for a person: once on the strategy the data points to, and again on the finished content before anything is final. We do the heavy reading and drafting. Your people sign off on the direction and on every word that goes out.
A layered compliance defense.
Before content is final it passes a multi-tier compliance check, not a single rule. Each tier catches a different class of risk, so a sensitive claim gets flagged and corrected early rather than slipping through to a published page.
Every byline is a real person.
Content attributed to a person or your brand goes to that named expert for final review before it publishes. We draft. Your expert approves. We never impersonate anyone, and nothing goes out under a name without that person's sign-off.
Then measure again
One team runs the whole loop, every month.
Most tools hand you a dashboard and leave the hard part, the content and the follow-through, to you. That is where momentum dies. We run it end to end: the monthly benchmark, the content on your domain, placement across the off-domain sources AI pulls from, and the re-measurement that shows whether the needle moved. You stay in control, and nothing goes out until you approve it.
01 · Build evidence
Publish authoritative, structured content AI systems can reliably crawl, understand, and cite. Drafted at machine speed, approved by a human every time.
02 · Establish identity
Make the engines associate the brand with the correct company and domain — entity resolution, URL-verified.
03 · Win the comparison
Head-to-head recommendation wins on a locked prompt set against a constant competitor field. The leading indicator.
04 · Expand influence
Wins accumulate into broader referencing. Mentions, visibility, and conversions are intended to follow the preference.
The objective is not to increase mentions. The objective is to change which company the AI recommends first.
The scoreboard
Measured on locked prompts, month over month, every number verified against the client's own domain.
SMS & messaging
+480% wins
verified AI wins in 11 weeks: 5 → 29 of 80 queries
Soapbox Bulletin
From sixth place to the most recommended brand in its benchmark, beating SimpleTexting, Wonder Cave, Peerly, Twilio, and DirectSnd on identical prompts.
Read the case study
Wealth management · brand redacted for compliance
+300% wins
May to June (1 → 4), from a zero April baseline, across 3 of 4 platforms
An independent wealth manager
More than a dozen firms share the client's name inside the AI engines. URL-grounded measurement found the real story: win the name first, then the market.
Read the case study
How a win is counted: a first-place recommendation for the client's verified domain on one locked buyer-intent prompt and platform test, rerun on the same schedule. On locked prompts against a constant competitor field, recommendation wins rose after the content program. The pattern is directional evidence, rerun monthly to test whether it holds, and it can shift as models and retrieval systems change.
Can you trust what goes out?
Every piece is sourced, human-approved, and yours to sign off.
The content we publish in your name holds to the principles everything Trinzik builds on. Every claim traces to a real source; what cannot be sourced is cut, not invented. A person curates, edits, and approves every piece, and every byline belongs to a real person who signs off on what carries their name. We win recommendations on the strength of the evidence, never by gaming or fabricating anything.
Who does the AI recommend first in your market?
A consult shows you, live, how ChatGPT, Claude, Perplexity, Gemini, and Grok answer your buyers' questions today, who is winning them, and what we would publish to take the recommendation.