VisibilityPro finds why ChatGPT, Claude, Perplexity, Gemini and Google's AI answers don't recommend your pages — then generates the fix, ships it to your CMS, and verifies it after publish. One workflow, from audit to citation.
The moat is the Implementation Planner Agent: it detects your CMS, classifies the change, picks the safest path — Assist · Draft · PR — and gates publish behind approval with a rollback path. AEO checkers stop at the score. We ship the fix.
Prefer to look first? See the agent loop · get the Chrome extension.
vs. 4–8 hours of manual analyst time. This is the AI leverage. We track it on every engagement.
Search isn't ten blue links anymore — it's one synthesized answer with footnotes. The unit of optimization changed from page → passage, and the KPI changed from rank → share of model. Your tools should change with it.
Most tools stop at step 2. We close the loop — your inbox shows decisions, not dashboards.
Each agent gets a tight tool allowlist — 2–4 vetted skills, not a generic LLM agent with shell access. The orchestrator routes findings, deduplicates, and writes every decision to a hash-chained audit trail you can export.
Audit your own page, your competitor's page, a Perplexity answer, or a SERP — all without leaving the tab. Side panel for the report, on-page overlay for the issues, chat for the questions.
The agent highlights the exact paragraph that's costing you a citation, with the rewritten version one click away.
Open a Perplexity answer, hit "Audit" — see which competitors are cited, which sources, and what your page is missing.
Select any text on the page: "Make answer-first", "Add FAQ pair", "Rewrite in brand voice", "Generate dev brief".
The extension never writes back to the live page. Drafts go to your CMS staging path; you approve every publish.
We help ambitious companies clarify positioning, sharpen narrative, and build identity systems that hold up under scrutiny. Our work spans research, naming, and visual identity for clients across fintech, healthcare and consumer.
A typical engagement runs 6–12 weeks, depending on scope and stakeholder count. We work in tight, accountable phases with clear deliverables at each step.
The same auditor your team uses in the browser is exposed as a Model Context Protocol server (stdio + HTTP), so your own agents — Claude, an internal copilot, a CI job — can call it directly. AI visibility becomes a tool other tools can invoke, not a dashboard a human has to open.
audit_page, then poll get_audit_report / list_audits. Pull the JSON-LD with generate_schema or diff a rival with compare_with_competitor.
check_ai_visibility returns measured monitoring results — share of model across the six engines — straight into your agent's context.
generate_fix and create_cms_draft produce a draft, never a publish. The backend independently verifies the fix was approved in the dashboard — an unapproved draft is refused server-side. MCP doesn't widen the blast radius.
Every CMS is different. The planner picks the safest path for yours — read-only by default, draft-first, publish only after explicit approval.
For Squarespace, Wix, and locked-down setups. The agent produces the exact paragraph, schema block, or alt-text — you paste it. Browser overlay shows where.
WordPress, Shopify, Webflow, Drupal, Magento, Joomla — the agent creates a draft revision with the diff. You preview, you publish. Rollback in one click.
Next, Astro, Nuxt, Hugo, Gatsby, headless CMSes. The agent opens a branch, commits, and files a PR with rationale + before/after diff for the reviewer.
Agentic tools should fail loud and small, not silent and big. Every action is gated, vetted, logged, and reversible.
Each agent gets 2–4 vetted skills, never a generic LLM with shell access. Community skills go through azhua-skill-vetter before they touch your data.
The default permission is read_only + draft_write. Publish, theme edits, robots.txt and redirects are gated by explicit user approval.
Pages can lie. Every fetched DOM is screened by lieutenant for injection patterns and tool-use jailbreaks before agents reason on it.
Every decision, generated fix, and CMS action is hash-chained via agent-audit-trail. Replay any audit, export to compliance, rollback any change.
The agent does the work. We own the outcome. Every escalation lands in a queue a human operator is actively watching.
Every escalation lands in our shared inbox. During business hours, a member of the four-person team picks it up — same people who built the agent, same people who run client delivery.
The agent does the work. We own the result. If the visibility score doesn't move on a Done-For-You engagement, we keep working until it does — no separate retainer, no scope renegotiation.
Approval friction, ambiguous fixes, YMYL boundaries — these stop the agent and hand back to the operator. You see the reason, the suggested next step, and the option to override or pause.
Approval is always the client's click — never the agent's.
Open any competitor page in your browser, hit "Compare with competitor" in the extension — Peeksy audits both pages side-by-side and shows you precisely what they have that you don't.
Heading hierarchy, answer-first sections, FAQ depth, cost / process / eligibility coverage — Peeksy surfaces the exact paragraphs giving competitors the citation edge over you.
JSON-LD types, missing FAQPage schema, author bios, trust signals, publisher entity links — the elements AI engines rely on when deciding who to cite in an answer.
Probe ChatGPT, Claude, Perplexity, Gemini, Google AI Overviews, and Google SERP on the same prompt set. See citation frequency, rank, and the exact answer snippets your competitor already owns.
One price per engagement — audit, fix generation, CMS implementation, post-publish verification, and a proven AI Visibility Score Delta. No subscription, no per-seat math. Itemised modifiers for YMYL, custom integrations and extra FAQs are listed below the matrix.
Brand discovery & scope · 10-agent AI-visibility audit · fix generation · CMS draft or PR · post-publish verification · AI Visibility Score Delta proof (ADR-022) · up to 50 FAQs generated.
Formal attestations (SOC 2, ISO 27001) are on the roadmap, tracked honestly on our security page — we don't badge certifications we don't hold.
21 active engagements. Public case studies publish after each client's first 30-day AI Visibility Score Delta lands (our North Star Metric) — we'd rather show the score moving than a quote that hasn't been earned yet.
Six build streams, one v1 release. Spec §13 + §17 enumerate the full feature set; each stream below ships behind a feature flag during build and is enabled at launch.
Per ADR-016, all six streams ship in the v1 release. The build is sequenced for engineering execution; the user-facing product is a single coherent launch — checker → fixer → monitor → multi-CMS — not a staged version ladder.
Spec §17. Each stream ships when the prior stream is stable in production.
Persistence layer · Stripe Usage Records metering · audit core — the engineering substrate.
Draft creation · FAQPage schema · preview links · post-publish verification.
Projects · client reports · white-label · team seats · approval links.
Prompt sets · AI visibility tracking · citation tracking · alerts.
Shopify · Webflow · GitHub PR · Drupal · Magento · Squarespace/Wix assist mode.
Mobile companion (approvals, notifications, weekly reports) is on the roadmap after Stream 6.
Book an engagement — we run the audit, generate the fixes, ship them to your CMS, and prove your AI Visibility Score Delta moved. Want to look first? Run a free demo audit, no card.
Not ready today? Remind me in 30 days.
Not a faceless tool. The four operators who built the agent stack are the same people who run your engagement and own the outcome.