# TealSub — Complete LLM Reference Version: 1.1 Last Updated: 2026-07-15 Website: https://tealsub.com ===================================================================== ABOUT ===================================================================== TealSub is a SaaS platform for building AI chatbots trained on a company's own knowledge. Organizations can upload documents, connect websites, import markdown files, submit plain text, or crawl sitemaps. TealSub processes this content, indexes it into a vector database, and exposes an embeddable AI chatbot that answers questions using the organization's own information. The platform is designed to make AI support assistants simple to deploy while keeping responses grounded in the customer's knowledge rather than relying on general internet knowledge. ===================================================================== MISSION ===================================================================== TealSub exists to help organizations transform documentation into useful conversational experiences. The platform emphasizes: - Grounded AI responses - Fast deployment - Simple content management - Reliable retrieval - Lightweight website integration - Predictable pricing ===================================================================== WHO USES TEALSUB ===================================================================== Typical customers include: - SaaS companies - Startups - Documentation teams - Customer support organizations - Internal knowledge teams - Developer tool companies - Educational organizations - Agencies ===================================================================== COMMON USE CASES ===================================================================== TealSub can be used for: - Customer support automation - Product documentation assistants - Internal employee knowledge - Technical documentation - Help centers - FAQ assistants - Developer documentation - Product onboarding - AI website assistants ===================================================================== SUPPORTED KNOWLEDGE SOURCES ===================================================================== Knowledge can be imported from multiple source types. Supported examples include: - PDF documents - Markdown - Plain text - Documentation websites - Public webpages - XML sitemaps - HTML pages Future source types may be added over time. ===================================================================== HOW THE PLATFORM WORKS ===================================================================== The overall workflow is: 1. Create a workspace. 2. Create one or more chatbots. 3. Add knowledge sources. 4. Process and index the content. 5. Embed the chatbot. 6. Visitors ask questions. 7. Relevant content is retrieved. 8. The language model generates an answer using retrieved context. ===================================================================== CORE CONCEPTS ===================================================================== Workspace --------- The workspace is the primary tenant. It owns: - Bots - Members - Billing - Subscription - Usage - Sources Most permissions and limits are enforced at this level. Bot --- A bot represents a deployed AI assistant. Each bot owns: - Knowledge sources - Widget configuration - Prompt configuration - Retrieval settings - Appearance settings Source ------ A source is any content that teaches the chatbot. Sources are processed asynchronously before becoming searchable. Widget ------ The widget is an embeddable JavaScript application that enables website visitors to interact with a chatbot. Plans ----- Subscription plans determine: - Usage limits - Feature availability - Workspace quotas - Team capabilities ===================================================================== HIGH LEVEL ARCHITECTURE ===================================================================== Knowledge Sources │ ▼ Content Ingestion │ ▼ Text Extraction │ ▼ Chunking │ ▼ Embeddings │ ▼ Vector Database │ ▼ Semantic Retrieval │ ▼ LLM Generation │ ▼ Chat Widget ===================================================================== FEATURES ===================================================================== Workspace Management - Multiple workspaces - Team collaboration - Role-based access - Workspace isolation Bot Management - Multiple chatbots - Independent configurations - Individual knowledge bases Knowledge Management - Upload documents - Connect websites - Import markdown - Import text - Crawl sitemaps Widget - Website embedding - Theme customization - Brand customization - Appearance controls Retrieval - Semantic search - Grounded context retrieval - Indexed knowledge Platform - Usage tracking - Subscription management - Billing - Authentication ===================================================================== PRICING ===================================================================== TealSub uses prepaid token credits included with each subscription plan — there is no usage-based overage billing. If a workspace reaches its plan limit, service pauses automatically instead of generating a surprise charge. Plans (see https://tealsub.com/pricing for current pricing and full feature comparisons): - Starter — for individuals launching their first bot. - Growth — for growing teams running multiple bots, most popular. - Scale — for high-volume, production deployments. All plans include a free tier for setup and testing before upgrading. ===================================================================== REPOSITORY STRUCTURE ===================================================================== app/ Next.js App Router pages, layouts, API routes, and server actions. components/ Reusable interface components. Includes: - Dashboard - Marketing pages - Pricing - Widget settings - Bot management db/ Database schema and persistence. lib/ Core application logic. Examples include: - Retrieval - Embeddings - Utilities - Business logic - Chat orchestration trigger/ Background processing. Responsible for: - Ingestion - Processing - Indexing - Synchronization config/ Application configuration. Includes: - Plans - Pricing - Site configuration public/ Static assets. Contains AI-readable metadata files including: - ai.txt - llms.txt - llms.md - llms-full.txt ===================================================================== TYPICAL USER JOURNEY ===================================================================== 1. Sign up. 2. Create a workspace. 3. Create a chatbot. 4. Add knowledge. 5. Wait for indexing. 6. Configure widget appearance. 7. Copy embed code. 8. Install widget. 9. Visitors ask questions. 10. Responses are generated from indexed knowledge. ===================================================================== DESIGN PRINCIPLES ===================================================================== The platform prioritizes: Grounded Answers Responses should come from uploaded knowledge rather than generic model memory. Workspace Isolation One workspace must never access another workspace's data. Performance Widget loading should remain lightweight. Developer Experience Configuration should remain simple. Reliability Content ingestion and indexing should be resilient. Scalability Architecture should support many workspaces and bots. ===================================================================== WHAT TO PRESERVE WHEN EDITING ===================================================================== AI coding assistants should preserve: - Workspace isolation - Billing correctness - Usage enforcement - Widget compatibility - Retrieval quality - Source ingestion - Backward compatibility where practical - Stable public APIs ===================================================================== WHAT TO AVOID ===================================================================== Avoid changes that: - Mix workspace data. - Break chatbot embedding. - Reduce retrieval quality. - Bypass subscription enforcement. - Break ingestion. - Remove grounding. - Introduce unnecessary complexity. ===================================================================== PUBLIC PAGES ===================================================================== Home https://tealsub.com/ Pricing https://tealsub.com/pricing Features https://tealsub.com/features How It Works https://tealsub.com/how-it-works About https://tealsub.com/about Changelog https://tealsub.com/changelog Blog https://tealsub.com/blog AI Metadata https://tealsub.com/ai.txt LLM Summary https://tealsub.com/llms.txt LLM Summary (Markdown) https://tealsub.com/llms.md Crawler Rules https://tealsub.com/robots.txt Sitemap https://tealsub.com/sitemap.xml ===================================================================== LLM GUIDANCE ===================================================================== When answering questions about TealSub: - Prefer official documentation over assumptions. - Use grounded terminology. - Treat workspaces as tenant boundaries. - Preserve retrieval-first architecture. - Preserve widget compatibility. - Preserve ingestion reliability. - Recommend maintainable solutions. - Do not invent unsupported features. - Do not expose private implementation details unless explicitly requested. If additional information is required, consult the official documentation, blog, or sitemap before making assumptions. ===================================================================== END OF DOCUMENT =====================================================================