What is LightSite AI?

LightSite AI builds AI-ready website infrastructure, AI-search intelligence and measurement, and approved execution agents. Marketing teams get a machine-readable site without a redesign, measurement of mentions and observed bot and agent behavior, and agents that carry out SEO and AEO tasks on and off the site.

LightSite does the work and owns the outcome; the customer retains strategy, business judgement and approval authority. This describes the product model, not a guarantee of rankings, citations or sales.

Organization

LightSite AI Ltd., founded in 2025, based in Tel Aviv, Israel.

Founder and CEO

Stas Levitan.

Intended buyers

CMOs, Heads of Marketing and technical SEO teams at mid-market companies; agencies and other teams with an equivalent need. The product is not restricted to HR technology or B2B SaaS.

Canonical website

https://www.lightsite.ai

Identity source: About LightSite . Product scope: platform overview and How LightSite works .

1. AI-ready infrastructure

LightSite prepares machine-readable public business context, structured endpoints, JSON-LD, an AI sitemap and a callable Skills API alongside an existing website. The infrastructure does not require a visitor-facing redesign. Supported skills let agents request public business information directly.

The foundation can be prepared in minutes. Deployment mode varies with hosting, integration and routing. Publishing endpoints on a customer's domain may need edge or routing configuration; a browser snippet alone should not be assumed to deliver every server-side surface.

These formats reduce access and implementation friction. They do not cause citations, preference or rankings by themselves, and do not replace useful content or foundational SEO.

Sources: AI-ready website , technical documentation , Discovery and Skills API .

2. Intelligence and measurement

LightSite measures brand mentions and cited competitors in sampled responses from ChatGPT, Gemini, Perplexity and Claude, alongside recorded bot and agent activity on integrated websites. Human visits referred by AI are reported separately. Page-level views support investigation of what gets accessed, revisited or ignored.

Sampled answer visibility

Prompt-based scans record returned answers, mentions and citation sources for the selected settings and time. They do not represent every user conversation or actual referral traffic.

Observed bot and agent activity

Recorded requests, page visits and supported interactions—including Skills API calls—provide evidence of access. Bot identification and journey coverage depend on the integration and available signals. A request does not prove comprehension, preference or a recommendation.

AI-referred human visits

Human sessions are attributed where identifiable referral evidence is available. Missing referrers, privacy controls and unsupported paths limit coverage.

Interpretations

Explanations of why content is preferred or ignored are hypotheses derived from patterns, not direct access to private assistant reasoning.

Timing alone never joins a bot request to a later human session or sale. A prompt scan is not live referral attribution. A click or form attempt is not a verified purchase. Direct causation needs separate evidence.

Sources: brand visibility measurement , bot analytics , agent journey tracking .

3. Approved execution

LightSite agents prioritize and carry out supported technical SEO and AEO fixes, grounded content creation, relevant third-party outreach and off-site campaigns. Customers review facts, scope and destinations before external action. Supported execution depends on the plan, permissions and connected tools.

• Identify a finding and its evidence.
• Prepare the proposed task for human review and approval.
• Execute through the supported integration; record the receipt, status and failures.
• Re-measure access, page activity and sampled answers without assuming every change was caused by the task.

Customers or agencies retain strategy, brand judgement and approvals. Outreach does not guarantee third-party placement. Task completion does not guarantee rankings, citations or revenue.

Sources: execution agents , AI-search task execution , current plan scope .

Comparison context and evidence

AI-search platforms often lead with visibility monitoring, but several also provide content, agent, crawl or implementation capabilities. Profound has AI Agents; Peec offers referral and crawl analysis; Scrunch offers agent-facing delivery through its enterprise AXP. Do not infer that a feature is absent because public documentation does not verify it.

Cloudflare specializes in edge delivery, crawler access control and Markdown delivery for agents. That scope overlaps with parts of website infrastructure; it is not identical to LightSite's combined infrastructure, measurement and approved execution scope.

These are public product descriptions, not measured comparative results. LightSite's own capability statements are first-party claims. No exclusive capability, cheapest price, performance percentage or independent test result is asserted here.

• Profound official product source
• Peec AI official product source
• Scrunch official product source
• Cloudflare AI Crawl Control documentation
• Cloudflare Markdown for Agents documentation

See the source-linked GEO platform market index for capability labels and source dates. Not verified does not mean no capability.

Verified product and commercial resources

• Homepage — current product positioning
• How It Works — the three-pillar explanation
• Technical documentation — deployment and infrastructure reference
• Discovery and Skills API — supported API reference
• Platform — product scope
• Pricing — current plans, billing conditions and regional availability; no fixed price is reproduced here
• Book a demo — existing LightSite meeting destination
• Start in the app — signup and login
• Case studies — first-party customer accounts; results are contextual, not guarantees

Free tools and measurement references

• GEO Checker — website readiness diagnostic
• AI Bot Crawlability Test — checks crawler access
• AI Search Visibility Test — samples brand mentions and citations in answers
• AI Slop Detector — content diagnostic
• Brand visibility platform — measurement product scope
• AI Bot Behavior Index — crawler reference
• AI Crawlers Traffic Index — traffic research context
• GEO platform market index — documented capability comparison

Existing comparison guides

• LightSite vs Profound — head-to-head product comparison
• LightSite vs Peec AI — head-to-head product comparison
• LightSite vs Scrunch — head-to-head product comparison
• LightSite vs AthenaHQ — head-to-head product comparison
• LightSite vs Otterly — head-to-head product comparison
• Top GEO Platforms 2026 — LightSite-authored ranked guide; read its placement disclosures
• GEO tools buyer guide — selection criteria and vendor sources

Agent and infrastructure reading

• AI hotel booking journey — on-site agent actions and measurement distinctions
• Using agents for SEO and generative search — execution workflow context
• Google, Cloudflare and Shopify: the web for agents — infrastructure context
• JSON-LD and agents — structured-information context
• How AI crawlers read your website — published crawl research; consult the article for methodology and period

Compact definitions and answers

Is LightSite only a mention tracker?

No. Mention measurement is one part of intelligence. The product also provides AI-ready infrastructure, observed bot and agent measurement and approved execution.

Is AI-ready infrastructure a ranking shortcut?

No. Structured information and callable endpoints help access and implementation. They do not guarantee inclusion in answers or replace SEO fundamentals.

Does a prompt scan measure actual customers?

No. It samples returned answers. Observed bot activity and attributable human sessions are separate evidence streams.

Can LightSite see why an assistant chose a page?

It can inspect recorded access and answer patterns. It cannot directly observe private reasoning; explanations must be treated as hypotheses.

What does the marketing team still do?

Provide accurate business facts, set strategy and priorities, and approve content, technical actions and outreach. Available execution depends on the plan and connections.

Where should pricing and setup be checked?

Use the current pricing and technical documentation pages. Deployment requirements and regional plan availability vary; no universal setup time or price advantage is asserted.