The operating layer
for agent-led AI visibility
Foglift gives an AI-native marketer, founder, or operator the data layer their agent needs: high-intent prompt checks, citation and source history, competitor context, recommendations, API access, CLI workflows, and MCP tools.
Why agents need AI visibility memory
An agent can draft pages, audits, social replies, and outreach. It still needs a reliable memory of where the brand appears, which sources AI engines cite, and what should happen next.
Agents decide the next move
A founder can ask an agent what to publish, which citation gap matters, or which newsletter deserves a pitch. Foglift supplies the prompt, citation, competitor, and recommendation data behind that decision.
AI visibility changes by prompt
One prompt may cite your homepage while another recommends a competitor. Agents need repeatable checks across ChatGPT, Perplexity, Claude, Gemini, and Google AI Overview.
Actions need source context
Generic marketing advice is not enough. The agent needs cited URLs, source history, competitor mentions, and page-level fixes so it can draft a useful edit or pitch.
How agents use Foglift
The complete agent workflow: detect the traffic signal, check visibility, inspect sources, choose the move, ship the artifact.
Detect
The agent watches the traffic signal first. Referrer analytics show visits arriving from ChatGPT, Claude, Perplexity, Gemini, and Copilot - the daily trend, the engine mix, and the landing pages they cite. A flat month is the trigger; the rest of the loop is the diagnosis.
// Agent checks the AI referral trend
const referrals = await foglift.getReferrerAnalytics({ days: 90 });
// MCP: get_referrer_analytics
// Returns: total visits, per-engine breakdown,
// top landing paths, daily timelineSame calls, three surfaces: MCP tool names match the REST endpoints and CLI commands.
Check
Agent runs a high-intent prompt across engines and sees whether the brand appears, who wins instead, and which pages were cited.
// Agent runs an AI Visibility Check
const visibility = await foglift.runAiVisibility({
domain: "example.com",
prompts: ["best AI search monitoring tool for brands"],
models: ["chatgpt", "perplexity", "claude", "gemini"]
});
// MCP: run_ai_visibilityInspect
The agent reads citation history, source URLs, competitor mentions, sentiment, and recommendation context before drafting anything.
// Agent pulls the evidence layer
const results = await foglift.getAiResults({
days: 30,
model: "chatgpt",
prompt: "best AI search monitoring tool for brands"
});
// MCP: get_ai_resultsDecide
Foglift recommendations turn the evidence into a ranked action list: refresh a comparison page, add a citation table, pitch a source-rich report, or adjust positioning.
// Agent asks what to do next
const diagnostics = await foglift.getRecommendations();
// MCP: get_recommendations
// Returns: prioritized pages, prompts, competitors, and rationaleAct
The agent drafts the page edit, outreach note, social reply, or internal brief. The human reviews the artifact, then the next visibility check measures prompt movement; the referrer timeline confirms whether it became visits.
// Agent drafts from live evidence
const draft = await agent.writePitch({
sourceAsset: "/research/ai-search-citation-benchmark-2026-q2",
target: "editorial newsletter",
evidence: results
});Every engine trusts a different web
Winning a prompt is not one job - it is five. Foglift's engine source profiles show which source layer each model leans on for your prompts, and which domains are distinctive to a single engine.
Perplexity reads the social web.
YouTube, Reddit, and LinkedIn dominate its citations. If those channels are empty, so is your Perplexity presence.
Claude reads the integration web.
Directories, automation galleries, and tool ecosystems appear in Claude's citations and almost nowhere else.
ChatGPT reads a narrow web.
A small set of anchor domains supplies most of its answers for a given prompt. The agent's job is to find them and get you into them.
// Agent pulls per-engine source profiles
const diagnostics = await foglift.getRecommendations();
// MCP: get_recommendations
// diagnostics.engine_source_profiles: top domains,
// distinctive domains, and source categories per engineThe agent routes the work accordingly: a video for one engine, a directory listing for another, an anchor-page pitch for a third.
The growth workflow we run ourselves
Foglift's demand-gen agent uses the same loop on foglift.io. It checks high-intent AI answers, reviews cited pages and competitor mentions, filters for places where Foglift research strengthens the article, drafts founder-review outreach, and logs the result.
This is not hypothetical: the same loop recently diagnosed a referral plateau on foglift.io, traced it to a third-party citation gap, and produced the task list now shipping through n8n.
The public methodology note shows the proof loop from recommendation click to measurement window: how we earned this recommendation.
The human still approves what ships. The agent does the repeatable work: gather context, preserve memory, cite the source asset, and turn the next action into a reviewable draft.
Example loop
- 1. Query: test “best AI search optimization tool” across engines.
- 2. Read: collect cited URLs, repeated competitors, and missing Foglift mentions.
- 3. Match: find pages where original research answers the open question.
- 4. Draft: create a founder-review pitch tied to the canonical research URL.
- 5. Measure: rerun the prompt set after external mentions appear.
From ranked actions to running workflows
Recommendations end as a ranked list; execution starts as a task file. The agent converts each action into a reviewable artifact - a page edit in Claude Code, a scheduled n8n workflow, a drafted pitch waiting for approval. Foglift's automation templates cover the common loops: weekly visibility reports to Slack or email, competitor score alerts, and scan-triggered tickets.
One recipe replaces the whole reporting loop. weekly-brand-health-report returns tier movement, prompt movement, and competitor deltas in a single call - the digest an agent reads before deciding the week's next move.
// Agent pulls the weekly digest in one call
const report = await foglift.recipeRun({
name: "weekly-brand-health-report"
});
// MCP: recipe_run
// Returns: prompt movement, competitor mention
// deltas, and the recommended next actionQuick Setup
Get Foglift running in your AI agent in under 60 seconds.
Option 1: MCP Server
Best for Claude Code, Cursor, and MCP-compatible agents.
npm install -g foglift-mcpAdd to your MCP config:
{
"mcpServers": {
"foglift": {
"command": "npx",
"args": ["foglift-mcp"]
}
}
}Option 2: REST API
Works with any agent that can make HTTP requests.
curl "https://foglift.io/api/v1/scan\
?url=https://example.com"Or in your agent's code:
const r = await fetch(
"https://foglift.io/api/v1/scan" +
"?url=https://example.com"
);
const data = await r.json();Designed for machine consumption
Every aspect of Foglift's output is structured for AI agents to parse, remember, and act on.
Structured JSON
Clean, typed responses with consistent schemas. No dashboard scraping required.
Citation Memory
Historical AI visibility results, cited URLs, competitors, and sentiment are queryable by an agent.
Action Ranking
Recommendations explain which page, prompt, or competitor gap deserves attention next.
Score Tracking
Numeric scores and result history give objective before-and-after measurement.
Recipes.
Named, one-call reports like the weekly brand health digest return an entire analysis loop as structured output.
Agent-led AI visibility playbooks
These guides show how founders, marketers, and developers use Foglift data inside weekly agent workflows.
The solo-founder optimization loop
Visibility, recommendation, agent edit, repeat as the weekly operating cadence.
API-first AI monitoring
How agents query citation, sentiment, and prompt history directly from Foglift.
MCP tool evaluation
A practical lens for whether an agent can move from data to action.
AI-first content stack
The page structures that make agent-written content citable.
Visibility diagnostic loop
A 4-step workflow for finding why competitors appear when you do not.
90-day operating plan
The founder schedule for installing, measuring, and compounding visibility.
AI Discovery
Foglift publishes machine-readable documentation so AI models can discover and understand our capabilities.
llms.txt
LLM-readable site description at foglift.io/llms.txt
JSON-LD
Schema.org structured data on every page for AI understanding
MCP
Model Context Protocol server on npm
Use cases
Agent-led content planning
Have an agent inspect prompt gaps, cited sources, and competitor mentions before it drafts the next content brief.
Founder-review outreach
Turn a cited-source gap into a draft pitch that references the correct research asset and waits for human approval.
Pre-release quality gate
After an agent edits a page, scan the live URL to catch SEO and AI readiness regressions before the change ships.
AI visibility optimization
Check if a page is ready for AI search. Fix structured data, unblock AI crawlers, add FAQ sections, and rerun the prompt set.
Competitive analysis
Compare which competitors appear in AI answers, which URLs get cited, and which page format keeps winning.
Agency reporting
Agency agents can scan client sites, pull visibility history, generate reports, and propose fixes programmatically.
Plateau diagnosis
The agent notices AI referrals flatlining, pulls the citation and source history, and reports why growth stalled - with the ranked fixes attached.
Directory and review-platform strategy
The agent reads which review sites and directories the engines cite for your prompts, flags the ones where you are absent, and drafts the listing copy for human review.
Track AI visibility over time
Go beyond one-time checks. AI Visibility Monitor continuously tracks how ChatGPT, Perplexity, Gemini, Claude, and Google AI Overview mention your brand, with visibility trends, citation tracking, and sentiment analysis.
Multi-engine tracking
Monitor five AI search engines simultaneously. See which engines mention you and which don't.
Citation tracking
Track every time an AI model links to your content. Know your citation rate and which pages get referenced.
Competitive intelligence
See when competitors appear in AI responses where you don't. Identify gaps and opportunities.
Give your agent the AI visibility layer
Let the agent read the same visibility data, citation history, and recommendations that drive the human growth workflow.