Generative Engine Optimization

One platform for AI visibility.

Calibrated scores across every answer engine that matters. Feature-level reasoning for every prompt. A simulated, prioritized path to improve. All from a single system of record.

Start freeSee pricing

01Measure

Seven engines. Seven distinct indexes.

Lokrix AI measures your presence across every major answer engine simultaneously — treating each as the distinct retrieval system it actually is, not averaging them into mush.

Monte-Carlo sampling
Each prompt runs K times across temperature, persona, and locale — delivering a presence probability with a 95% confidence interval, never a single-run estimate that flatters or misleads.
Per-engine grounding
ChatGPT grounds on Bing (~87% citation match). Google AI Overviews and AI Mode both pull from Google — yet cite the same URL only ~14% of the time; Lokrix AI treats them as separate engines with separate scores.
Detection pipeline
URL-match + brand NER + semantic similarity — three signals that catch citations even without a direct domain link.

Source: Answer-engine citation-source analyses (2024–2026), summarised in CLAUDE.md §4

Citation source share, by engine

Where a published figure exists

Source: Answer-engine citation-source analyses (2024–2026), summarised in CLAUDE.md §4.

  • ChatGPT
  • Google AI Overviews
  • Google AI Mode
  • Perplexity
  • Gemini
  • Grok (beta)
  • Claude

02Reason

Feature-level attribution explains every score.

Lokrix AI attributes each score to the measurable factors that predict citation — so you know which lever to pull, and for which engine, before you spend a word.

SHAP-based decomposition
Each engine score is decomposed into the contribution of every content and technical feature — index rank, expert quotes, schema markup, freshness, and earned-media corroboration.
Engine-specific drivers
Reddit citations dominate for Perplexity; Bing rank for ChatGPT; schema markup for Gemini. Attribution is engine-specific — not one averaged signal.
Competitor attribution gap
See exactly how far ahead or behind you are on every factor for every tracked competitor — down to the individual engine and prompt cluster.
Feature-level attributionIllustrative
Bing top-10 rank
+22 pts
Strongest single predictor for ChatGPT citations (~87% match)
Expert-attributed quotes
+15 pts
Princeton GEO: +41% lift — highest-leverage on-page lever
Reddit earned media
+11 pts
~47% of Perplexity’s top citations come from Reddit
Schema.org markup
+7 pts
Gemini is schema-sensitive; structured data aids grounding

Source: Answer-engine citation-source analyses (2024–2026), summarised in CLAUDE.md §4; Princeton “GEO: Generative Engine Optimization” study (Aggarwal et al., KDD 2024).

03Optimize

A simulated path to #1 — not a guaranteed ranking.

The optimization engine ranks every actionable move by predicted visibility uplift, simulated from the calibrated model before you spend a word. Uplift is predicted — not promised.

Proven levers, prioritized
Expert quotations +41%, statistics +40%, authoritative citations +30%. These are the highest-leverage moves — ranked, per page, per engine.
No filler recommendations
llms.txt has no measured effect on AI citation rates in current research. You will never find it on a Lokrix AI recommendation list.
Agentic fix mode
Apply recommendations automatically via the MCP server — Lokrix AI acts as a tool inside your own AI workflows.

Source: Princeton “GEO: Generative Engine Optimization” study (Aggarwal et al., KDD 2024)

Citation lift by content lever

Relative to an unoptimised page

Source: Princeton “GEO: Generative Engine Optimization” study (Aggarwal et al., KDD 2024).

Note: llms.txt has no measured effect on AI citation rates. You will not find it on Lokrix AI’s recommendation list.

04Track

Catch visibility regressions before they compound.

Every scan adds a point to a longitudinal history with a 95% CI band. Regression alerts fire when your presence drops outside the expected range — so you know before anyone else does.

CI-bounded history
The trend line shows the honest 95% confidence interval as a shaded band. Provenance is badged (live / cached / predicted) on every data point.
Regression alerts
Email, Slack, and webhook notifications when presence drops — segmented by engine, prompt cluster, or competitor share-of-voice shift.
Full history retained
Every scan point is stored with its original 95% CI band and provenance badge — live, cached, or predicted — so you can track trends over time.
Illustrative

Surfaced score over time

With 95% CI band

Provenance is badged on every point — live, cached, or predicted. The band is the honest 95% CI, not decoration.

05Compete

Know where you stand against every alternative.

Track up to 25 competitors' share of voice across every engine and prompt cluster, updated every scan cycle — and see the exact attribution gaps that explain the delta.

Share-of-voice per prompt
For each tracked prompt, see which competitors were cited, with what probability, and with what confidence interval — across every engine simultaneously.
Attribution gap
Identify the specific content and technical factors where competitors outperform you, so optimization effort goes to the highest-value gaps first.
Competitor benchmarking
Every scan includes competitor share-of-voice benchmarking. For high-volume or bespoke benchmarking setups, contact us for a custom quote.
Share of voice — “best CRM software”Illustrative
Category leader
91%
Rival A
78%
Your site
63%
Rival B
54%
Rival C
43%

06Automate

A full REST API and MCP server for your own agents.

Every Lokrix AI capability — scanning, scoring, optimization, history — is accessible programmatically via a public REST API and an integrated MCP server.

REST API
POST /v1/scans to trigger a scan; GET /v1/scans/:id to poll for provenance-tagged scores with confidence intervals. API keys are per-tenant, per-scope.
MCP tools
lokrix.analyze_url and lokrix.score_prompt let Claude, GPT-4o, and Gemini call Lokrix AI directly as a tool inside your own agentic workflows.
Multi-tenant isolation
Row-level security enforced at every query. RBAC (Owner / Admin / Editor / Viewer). Enterprise SSO / SAML / SCIM via WorkOS. SOC 2 readiness and GDPR / CCPA posture.
REST API + MCP tools
# REST API

POST /v1/scans
Content-Type: application/json

{ "url": "https://acme.com" }

→ 202 Accepted
{ "id": "scan_01HXYZ", "status": "pending" }

GET /v1/scans/:id

→ 200 OK
{
  "id": "scan_01HXYZ",
  "status": "done",
  "score": {
    "surfacedScore": 73,
    "ci": { "lo": 61, "hi": 84 },
    "provenance": "live"
  }
}

# MCP server (lokrix-mcp-server)

lokrix.analyze_url
  in:  { "url": "https://acme.com" }
  out: { "brand": "Acme",
         "universe": { "promptCount": 50 } }

lokrix.score_prompt
  in:  { "prompt": "best CRM software",
          "url": "https://acme.com" }
  out: { "presenceProbability": 0.73,
          "ci": { "lo": 0.61, "hi": 0.84 },
          "provenance": "live" }

Find out if AI answers name you.

Start free with 50 trial credits, then add a plan plus credits to unlock live multi-engine scans, history, competitor benchmarking, and the full optimization engine.

Start freeView pricing

Start free · 50 trial credits · no credit card · then a plan + credits.