AI Recommendation Intelligence Platform

See how AI sees your brand.

Intelligence for the AI decision layer.

Buyers now ask ChatGPT, Gemini and Perplexity before they ask you. Those answers already compare, rank and recommend brands in your category — shaping demand before anyone reaches your site. Radar37 measures your position in that decision layer, and shows you how to lead it.

Multi-signal frameworkEvery major AI engineBuilt with GEO & SEO practitioners
AI answer · monitored
Which platform should our enterprise shortlist?
The options most often recommended are Competitor A, Competitor B, and — for teams prioritising value and onboarding — your brand, which reviewers consistently rate highly.
Review · G2Community · RedditOwned · docs
Radar37 readingVisibility 76% · Position #1.6 · Sentiment +19
Illustrative reading
01
The shift

Google decided who got discovered.
AI decides who gets recommended.

A generation of growth was built on ranking in search. The next will be built on being the answer. This transition is already underway — and it changes who customers ever consider.

Past · the search era

You optimised for Google.

Buyers searched, scanned a page of links, and chose. Visibility was a ranking you could measure and influence.

Present · the ask era

Buyers ask AI first.

They put the question to an assistant and act on one synthesised answer — often without a single click to your site.

Future · the recommendation layer

AI becomes the gatekeeper.

The model's recommendation is the shortlist. If you're not in it, you're not in the consideration set at all.

02
Why this matters

The risk is invisible — until the pipeline is already gone.

Because the answer layer is unmeasured, the damage compounds quietly. By the time it shows up in revenue, competitors have owned the recommendation for months.

Competitors become the default answer

If AI consistently names a rival first, it quietly becomes the category's preferred choice in every buyer's mind.

Category visibility erodes

Slip out of the answer for high-intent questions and you lose presence exactly where decisions are formed.

AI-driven demand leaks away

Interest that once reached you through search is now resolved inside the answer — demand you never see.

Declining influence in the journey

Your brand's narrative is increasingly written by sources you don't control — and you have no read on it.

The brands that measure the answer layer first will shape it. The rest will inherit whatever AI decides.

03
The framework

Six signals. One picture of your standing in AI.

Most tools count mentions. Radar37 evaluates how AI actually treats your brand — across six signals that map to real influence, not vanity metrics.

M01

Visibility

Share of all AI answers that mention the brand at least once.

76%
M02

Net Sentiment

Reader-impression score, −100 to +100. A neutral-toned scandal still reads negative.

+19
−1000+100
M03

Total Mentions

Raw count of valid mentions across the tracked set.

0 valid
M04

Average Position

Mean reading-order rank — counted only in genuinely comparative answers where competitors also appear. Otherwise null.

#1.6
M06

Signal

A verdict per prompt, from citation rate × platform spread × sentiment.

WinningWatchGap
M05

Share of Voice

Of all answers where competitors appear, how often the brand is one of them.

Rival A35%
You28%
Rival C21%
REF

Citation Source Mix

Where the answers that cite you come from — the credibility behind your visibility.

Review46%
Owned22%
Community18%
Figures shown are an illustrative readout
04
How AI decides

A recommendation is the end of a chain. We see all of it.

AI doesn't invent opinions — it assembles them from sources it trusts. Radar37 instruments every stage from raw source to customer decision, so influence becomes traceable.

Sources
Reviews, communities, press, your own content.
Citations
Which of those sources the model actually pulls.
AI models
ChatGPT, Gemini and Perplexity weigh and synthesise.
Recommendation
The brands the answer names, ranks and frames.
Customer decision
Who gets shortlisted, trusted and chosen.

Radar37 instruments the full chain — and shows you where to intervene to change the outcome.

05
Enterprise intelligence

Four layers of intelligence, not a dashboard of metrics.

Each signal rolls up into strategic intelligence your leadership can act on — competitive, source-level, market-wide, and decision-level.

Competitive intelligence

Know exactly who AI prefers, and why.

Track your share of voice and ranking against every rival, per prompt and per engine — and see the moment a competitor starts winning the recommendation.

Share of VoiceHead-to-head positionWin / loss by prompt
Source intelligence

Understand how AI forms its opinion.

See which sources — review platforms, communities, editorial, your own domains — shape the answers about you, so you invest where credibility is actually built.

9 source typesCitation shareSource × topic
Market intelligence

Read the whole category, not just yourself.

Watch how the answer space for your market moves over time — emerging topics, shifting sentiment, and where the category conversation is migrating.

Visibility over timeTopic trendsCategory drift
Recommendation intelligence

Turn every reading into a prioritised move.

Signal Detection grades each prompt Winning, Watch or Gap and resolves it into an action map — the fastest paths to becoming the recommended answer.

Winning / Watch / GapAction mapOutcome-ranked
06
The platform, in action

Built for the decision, not the dashboard.

Every view answers a leadership question: where do we stand, where are we losing, and what should we do next.

VISIBILITY

Your standing, at a glance

The share of AI answers in your category that include your brand.

76%
of category
answers include you
this period

Visibility over time

YouCategory
Catch a competitor reshaping the answer space the week it happens — not a quarter later.

Source × Topic — where the conversation lives

Citation density across topics and source types. Dark cells are where AI draws its answers — and where presence is decided.
OwnedReviewComm.EditorialExpert

The action map

What each reading tells you to do next.
GapCategory buying queries0% visibility · own + review missing
WatchPricing comparisonsranked #4 · lift review presence
WinningImplementation questions100% · positive · hold position
Illustrative action queue
07
Under the hood

Accuracy is an engineering problem. We built for it.

Every number on the dashboard is the output of a measurement pipeline — multi-engine, validated and calibrated. This is what runs behind a single reading.

01

Multi-engine orchestration

Every prompt is run across ChatGPT, Gemini and Perplexity. Presence is read per model, never assumed from one.

02

Journey-wide prompt sampling

Thousands of prompt variants mapped to real buying-journey intent — not a handful of vanity queries.

03

Token-level answer parsing

Each response is analysed at the token and citation level to detect mentions, reading-order position, and the sources behind them.

04

Multi-stage validation

Proper-noun match, brand-context alignment and hard filters strip false positives, refusals and clarifications before anything is counted.

05

Reader-impression sentiment

Scored by the impression left on a reader, not the writer's tone — a neutrally-written negative still reads negative.

06

Source classification & attribution

Every citation resolved into one of nine source types, first-match-wins, then attributed back to the brand and topic.

07

Calibrated precision

Sampling volume and frequency scale to the confidence a decision needs — with the statistical tradeoffs made explicit, not hidden.

Measurement spec
EnginesChatGPT · Gemini · Perplexity
Prompt coverageJourney-wide, thousands of variants
Runs / promptScaled to required precision
AnalysisToken & citation level
Signals6 — visibility → signal
Source types9 · first-match-wins
ValidationProper-noun + context + filter
SentimentReader-impression
RefreshPulse → continuous
08
Why Radar37

Methodology built by people who shape growth for category leaders.

Built with GEO & SEO practitioners

Shaped alongside experts who have driven growth for category-leading brands — the judgment of specialists, engraved into the methodology.

A multi-signal framework

Six signals — visibility, share of voice, position, sentiment, citations, detection — instead of a single mention count that flatters and misleads.

Precision calibrated to the decision

Sampling volume and tracking frequency scale to the certainty a decision demands — enterprise-grade rigour, not a one-off snapshot.

Future-ready by design

New engines and surfaces fold into the same framework, so your view of the AI decision layer stays complete as it evolves.

PRECISION SCALES TO NEED
Pulse
Direction
Standard
Decisions
Forensic
Defensible
Tiers illustrate the model — depth scales to need
09
Pricing

One dollar per prompt. That's the entire model.

No tiers to decode, no seat math. You pay only for what you measure — each tracked prompt, fully analysed across every engine.

Simple · usage-based
$1/ prompt measured

Every tracked prompt — run, parsed and graded end-to-end — billed as one predictable unit.

What a single prompt includes

A $1 prompt is a full measurement — not a single API call.
Run across every major AI engine
All six signals, fully computed
Multi-stage validation & false-positive filtering
Token-level parsing & citation classification
Source attribution & competitive position
Signal grade & recommended action

Each prompt is measured across every major engine, with multiple runs and full token-level analysis — the depth other tools wrap in platform fees, priced as one transparent unit. The most precise reading in the market, and the most cost-effective.

True intelligence begins with awareness. Radar37 brings it to the answer layer — where AI now decides who gets seen, trusted and chosen.

Own your position in the AI decision layer.

See exactly how AI evaluates, compares and recommends your brand — and get the plan to lead your category as discovery shifts to the answer.

See how ChatGPT, Gemini & Perplexity describe and recommend your brand  ·  free  ·  no credit card

YOU