Overview
GotCited is a complete AI search visibility platform, and you get the full source code. A brand adds its website. GotCited writes the buyer questions people ask AI assistants, sends them to every AI engine you have connected, and reads each answer to see whether the brand was named, who was named instead, and which sites were cited. Every rate comes with a 90% confidence interval, so a change between runs can be told apart from noise.
It does not stop at the report. Each problem becomes a fix: generated config ready to paste, an AI-drafted page or correction for approval, or a guided outreach task. After a fix ships, GotCited re-measures it against prompts that were held out as a control. It suits SEO and content agencies, B2B marketing teams, and operators who want to run it as their own SaaS. It comes with workspaces and teams, Stripe plans and credits, white-label client reports and an operator admin panel.
Features
Measurement
- Nine engine adapters: ChatGPT, Claude, Gemini, Perplexity and Grok through their APIs with web search; Google AI Overviews and AI Mode through DataForSEO; Copilot, Meta AI and the real ChatGPT app captured through Bright Data. Each engine switches on when its key is set.
- Buyer prompt generation: Prompts are written from the brand's site, category and competitors, tagged by buyer stage (discovery, comparison, decision) and persona. Branded prompts are reported separately.
- Control prompts: One in five prompts is held out and never optimised, so lift is measured against a real baseline.
- Answer extraction: Each answer is parsed for brand mentions, position, competitors named and cited URLs. Engine, model, capture method, market and time are stored on every answer.
- Statistics: Wilson 90% intervals on every rate, and a two-proportion test to flag significant changes between runs.
- Prompt demand: Monthly search volume from DataForSEO when connected, Search Console impressions when connected, otherwise a Low/Medium/High tier from the model.
- 46 markets: Prompts and answers in the buyer's language and country, from en-US to ja-JP and ar-SA.
- Scheduling: Weekly runs per brand, manual runs within plan limits, and a weekly email digest that says whether the change is significant.
Dashboard
- Overview: Mention rate gauge, a flow chart from buyer stage to engine to the brand named, share of voice, a citation map, next actions, prompts to win, lift vs control and recent activity.
- Visibility: Mention rate over time with the 90% band, shipped fixes marked on the chart, and a per-engine table.
- Prompts: Every tracked prompt with stage, demand and result per engine. Add prompts, generate 20 more, filter by stage and refresh demand.
- Sources: Every domain the engines cite, and a ranked short list of third-party sites (review sites, communities, publishers) where showing up would change the most answers.
- Command palette: Keyboard search across brands, pages and actions.
Fixes
- Site scan: Checks for robots.txt rules or a WAF blocking AI retrieval bots, an unreachable site, JavaScript-only content, missing Organization schema, missing meta description, missing sitemap and unverified Bing indexing.
- PageSpeed follow-up: Optional Lighthouse audit of the homepage for slow server response, indexability, page weight and SEO basics.
- Fix queue: Findings sorted by severity and impact, filtered by how they ship: auto-fix (generated robots.txt rules and JSON-LD to copy), draft for approval, or guided task.
- AI drafts: Answer-first pages for prompts the brand is losing, outreach pitches for source gaps, and correction requests for wrong facts, written in the market's language.
- Content plan: After each measurement, GotCited proposes the pages to publish and drafts them on request.
- Accuracy: Draft a fact sheet from the website, edit and save it, and every run checks AI answers against it for wrong prices, plans or claims.
- Status tracking: Mark fixes drafted, shipped or dismissed. Shipped fixes are re-measured against control prompts.
Traffic and crawlers
- AI referral traffic: GA4 sessions and key events from AI assistants by engine and week, and Search Console queries, synced daily.
- Pipeline: HubSpot deals whose contacts first arrived from an AI assistant.
- AI crawler logs: A per-brand ingest URL accepts JSON, NDJSON (Vercel log drains, Cloudflare Logpush, gzipped or not) and nginx/Apache combined logs. Shows which AI bots read the site, which pages, live fetches and blocked responses. Bots are verified against the operators' published IP ranges.
Reports, teams and integrations
- Client reports: Share links for the last 7, 30 or 90 days, which can be revoked, and a print-to-PDF layout.
- White-label: On the Agency plan, shared reports carry the agency's name, logo and accent colour, and can hide GotCited branding.
- Workspaces and teams: Several brands per workspace, several workspaces per user, and owner, admin and member roles with email invites.
- Slack: Alerts to a channel through OAuth, with per-event settings and a test message.
- REST API: Read brands, visibility, prompts, fixes, sources, facts, traffic, crawlers and runs, start a run and update fix status. Uses API keys and serves an OpenAPI document.
- MCP server: Ten tools for AI agents, including get_visibility, list_fixes, list_wrong_facts and start_measurement.
Accounts, billing and admin
- Auth: Email and password with email verification, password reset and security emails, built on Better Auth.
- Stripe billing: Free, Starter, Growth and Agency plans with limits on brands, prompts, engines, runs, drafts and seats. Checkout, customer portal and webhook sync.
- Credits: Credit packs pay for extra drafts, re-plans and manual runs on any plan. Credits do not expire.
- Admin panel: Platform overview, organizations, users, runs, a credit ledger, AI engine cost, a system page that shows which settings are present without showing values, and an audit log.
- Operator actions: Grant or remove credits, set a plan, pause or restore a workspace, rename an organization, and create a demo agency workspace.
- Model-agnostic LLM layer: Anthropic by default, with OpenAI, Google and DeepSeek as fallbacks, fast/balanced/deep tiers and a circuit breaker.
Marketing site
- Landing page: 3D hero, product film with captions, how-it-works, method, pricing, FAQ and a free-audit call to action.
- SEO basics: Sitemap, robots, llms.txt, generated Open Graph and Twitter images, privacy and terms pages.
Good to know
- Copilot, Meta AI and the ChatGPT app capture need your own Bright Data Web Scraper datasets. The landing page lists Copilot and Meta AI as "Soon", and these adapters were not tested for this listing.
- "Auto-fix" means GotCited generates the robots.txt rules and JSON-LD for you to copy. It does not write to your site, CMS or repository.
- The AI cost page counts engine answers only. Extraction, planning and drafting calls are not included yet.
- Background jobs run inside the web process. A long-running server is needed, or set
SCHEDULER=off and call /api/cron every 10 minutes.
- There is no social sign-in. Accounts use email and password.
Tech stack
- Next.js 16 (App Router, Turbopack), React 19, TypeScript, Tailwind CSS 4
- Better Auth (email and password)
- PostgreSQL 17 with Drizzle ORM
- Anthropic, OpenAI, Google Gemini, Perplexity and xAI APIs; DataForSEO and Bright Data
- Stripe, Resend, Zod
- Three.js (React Three Fiber), Motion, React Aria Components, d3-force
- pnpm workspaces, Node 22, Docker
What's included
- Full monorepo: the
apps/web app plus the core, db, engines and llm packages
- Drizzle schema, applied with
pnpm db:push
- Marketing site with the 68-second product film, captions and brand assets
- Product plan and brand guide (
PLAN.md, BRAND.md)
Dockerfile, docker-compose.yml for local Postgres, and railway.json
- Scripts to check which API keys work, seed a local admin user, and evaluate prompt and content quality
- Unit tests for the core package
.env.example documenting the main keys
Requirements and third-party costs
You need Node 22+, pnpm and a PostgreSQL database. The app runs background jobs in its own process, so it needs an always-on server or container (a Dockerfile is provided). Each measurement sends many prompts to paid AI APIs, so engine costs grow with brands, prompts and engines.
- Hosting and database (required): Any Node or Docker host plus Postgres, such as Railway, Coolify, Neon or Supabase.
- Anthropic API (required): Runs prompt generation, answer extraction and drafting, and the Claude engine.
- OpenAI, Gemini, Perplexity and xAI APIs (optional): Each turns on one more engine to measure. OpenAI, Gemini and DeepSeek can also act as fallback models.
- DataForSEO (optional): Google AI Overviews, Google AI Mode and search volume for prompt demand.
- Bright Data (optional): Copilot, Meta AI and ChatGPT app capture. Needs your own datasets.
- Resend (required in production): Verification, password reset, invites and weekly digests. In development, emails print to the server log.
- Stripe (optional): Needed to charge customers for plans and credits.
- Google Cloud OAuth app (optional): GA4 and Search Console import.
- HubSpot and Slack apps (optional): Pipeline attribution and alerts.
- Google PageSpeed Insights API key (optional): The homepage performance audit. It is free with a Google Cloud key.
The purchase covers the source code only. Service fees are billed by the providers listed above.