Lettertrace: Free AI Visibility Tracking with Your Own API Keys

Lettertrace is a free, open-source AI visibility monitoring tool that tracks how often brands appear in answers from ChatGPT, Claude, Gemini, Google AI Overviews, and Perplexity.

Add a brand, competitors, and topics to monitor. Lettertrace generates realistic question variations, sends them to the selected answer engines, records the responses and cited sources, then tracks visibility, share of voice, prominence, sentiment, and recommendations across repeated runs.

Visit Lettertrace

Lettertrace GitHub Repo

How Lettertrace Tracks AI Visibility

Lettertrace organizes monitoring around brands, competitors, topics, prompts, and runs. A project stores the brand name, aliases, domain, model settings, and schedule. Topics describe the search areas that matter to the business, while prompt variations represent questions users might ask an AI assistant.

Each monitoring run sends the active prompts to the selected answer engine and stores the returned text. Brand detection starts with case-insensitive matching against the configured brand and competitor names and aliases. Mentioned entities then receive sentiment and recommendation classifications. The dashboard aggregates the results into visibility rates, share of voice, prominence, and trend data.

Key Features

Queries Claude, ChatGPT, Gemini, Google AI Overviews, and Perplexity Sonar with forced web search on each supported engine.
Generates prompt variations for each topic from a plain description of what buyers ask.
Detects brand and competitor mentions through word-boundary name and alias matching, then classifies sentiment and recommendation status with a second model call.
Tracks visibility, share of voice, prominence, and sentiment across runs and stores every result for trend charts.
Benchmarks named competitors and surfaces “discovered” competitors that answers already named but nobody tracks yet.
Runs multiple brand organizations from one account and switches between them from the sidebar.
Captures the exact sources a model cites during web search, including whether the brand’s own pages appear.
Schedules daily or weekly runs through a cron endpoint, billed to the account owner’s own key.

How to Use Lettertrace

1. Create an account at Lettertrace with email, Google, or GitHub authentication.

2. Open Settings and add an Anthropic, OpenAI, Google, or Perplexity API key for the answer engines you plan to monitor.

3. Create the brand project with its name, domain, and aliases.

4. Add the competitors that belong in the share-of-voice comparison.

5. Create a topic and generate prompt variations. Review the prompts before running a large monitoring set.

6. Start a run and open the results for brand visibility, share of voice, sentiment, citations, and topic-level results.

7. Set a daily or weekly schedule after the prompt set and provider spending level match the monitoring plan.

Self-Hosting Lettertrace

The self-hosted setup needs a Supabase project, the included database schema, provider credentials, and an encryption key for stored BYOK credentials. Node 22 LTS is the recommended runtime.

git clone https://github.com/letterstory/lettertrace.git
cd lettertrace
cp .env.example .env.local
npm install
npm run dev

Apply supabase/schema.sql to the Supabase project before using the application, then add the required Supabase values and ENCRYPTION_KEY to .env.local. Scheduled runs also need the service-role credential and CRON_SECRET.