# Kelu

> Kelu is an AI knowledge assistant. It reads your docs, help center, past tickets, repos and wikis into one knowledge base, then answers customers and your team on your website, in your helpdesk (Zendesk, Intercom, Freshdesk, Front, Linear), in Slack and Microsoft Teams, in AI coding tools through MCP, and through a REST API. Every answer from your docs links to its sources; when the sources don't cover a question, Kelu says it doesn't know.

Key facts:
- Plans: Free (1 knowledge base, 500 indexed pages, 100 AI questions/month), Pro ($99/month or $79/month billed yearly; 14-day free trial), Enterprise (custom). Details: https://www.kelu.dev/pricing/
- Security: credentials encrypted with AES-256-GCM; PII masking and zero retention available on every plan; SSO (OIDC/SAML) on Enterprise; SOC 2 Type II audit in progress (no report yet); data stored in the United States.
- Contact: sales@kelu.dev, support@kelu.dev, security@kelu.dev

## Product
- [Pricing: Free, Pro and Enterprise Plans](https://www.kelu.dev/pricing/): Start free with no credit card. Pro is $99 a month and adds Slack and Teams bots and connectors. Enterprise adds SSO, unlimited usage and a custom SLA.
- [FAQ: RAG, Citations, Security and Pricing](https://www.kelu.dev/faq/): Frequently asked questions about Kelu — RAG, citations, Support Ticket AI, browser extension, MCP, BYOK, security, pricing, and quota enforcement.
- [Documentation: Quickstart, Guides and API Reference](https://www.kelu.dev/docs/): Learn how to use Kelu: connect knowledge sources, embed the AI chat widget, call the chat and search APIs, and read your analytics. Start with the quickstart.
- [Quickstart: Your First AI Answer in 5 Minutes](https://www.kelu.dev/docs/quickstart/): Get your first AI answer in under 5 minutes: create a free knowledge base, add your docs URL, let Kelu crawl it, then embed the widget on your site.
- [API Reference: Chat, Search and MCP Endpoints](https://www.kelu.dev/docs/api/): Kelu public API reference: client-key auth, streaming chat with citations, hybrid search, form deflection, WebSocket chat and the MCP server.
- [Security: Encryption, SSO, PII Masking and GDPR](https://www.kelu.dev/security/): How Kelu protects your data: AES-256-GCM encrypted credentials, SSO on Enterprise, PII masking and zero retention on every plan, a GDPR DPA, SOC 2 in progress.
- [Trust Center: Security, Privacy and Compliance](https://www.kelu.dev/trust/): Kelu's security and privacy checklist: workspace isolation, encryption, PII masking, zero retention, no training on your data, SSO, audit logs and a DPA.
- [Changelog: New Features and Product Updates](https://www.kelu.dev/changelog/): See what's new in Kelu: new features, connectors and fixes, listed by release with dates.
- [About Kelu: AI Answers From Your Company's Knowledge](https://www.kelu.dev/about/): Learn about Kelu's mission to make every enterprise's knowledge instantly accessible through accurate AI answers with citations.

## Features
- [AI Chat with Citations from Your Docs](https://www.kelu.dev/features/ai-chat/): Retrieval-augmented generation with source citations, follow-up questions, and conversation memory. Every answer is grounded in your company's knowledge.
- [Analytics for Your Knowledge Base](https://www.kelu.dev/features/analytics/): Self-serve rate by intent, unanswered gaps, feedback, citation coverage and latency in traces. Know what your company knowledge is missing.
- [API and SDKs: Build on the Answer Pipeline](https://www.kelu.dev/features/api-sdks/): REST API, widget script and a per-knowledge-base MCP server for streaming AI chat, search and feedback, plus React and JS SDKs on npm.
- [Browser Extension: Your Knowledge in a Side Panel](https://www.kelu.dev/features/browser-extension/): Kelu's extension for Chrome, Edge and Opera opens your knowledge base in the browser side panel, so you can ask about the page you're on without switching tabs.
- [Bring Your Own Model: Use Your Own LLM Key](https://www.kelu.dev/features/byok/): Run Kelu on your own OpenAI, Anthropic, Google Gemini or AWS Bedrock account. You keep the model choice, the usage bill and your existing data agreement.
- [Citations and Grounding: Answers That Show Sources](https://www.kelu.dev/features/citations/): Every Kelu answer from your docs carries numbered citations to the exact source. If your docs don't cover a question, it says so.
- [Doc Proposals: Drafts for Unanswered Questions](https://www.kelu.dev/features/doc-proposals/): Kelu finds the questions your docs can't answer well, drafts the missing answer, and hands the draft to your team to review and publish.
- [Answer Quality: Find and Fix Weak Answers](https://www.kelu.dev/features/evals-quality/): Kelu says when your docs don't cover a question, collects user feedback, records full traces, and turns weak spots into a clear content backlog you can act on.
- [Hybrid Search: Meaning and Keywords Together](https://www.kelu.dev/features/hybrid-search/): Kelu blends semantic vector search with full-text ranking, so a question finds the right page even in different words, and exact error codes still match.
- [Knowledge Sources: Connect and Sync Your Docs](https://www.kelu.dev/features/knowledge-sources/): Connect docs sites, Confluence, Notion, Google Drive, OneDrive, GitHub, help centers and more. Kelu keeps each one in sync and answers from it with citations.
- [MCP Server — Your docs, inside every AI coding tool](https://www.kelu.dev/features/mcp-server/): Every Kelu knowledge base is also an MCP server. Cursor, Claude and other MCP clients can search your current docs instead of relying on old training data.
- [PII Masking and Zero Retention for AI Answers](https://www.kelu.dev/features/pii-and-zero-retention/): Mask emails, phone numbers, card numbers and other PII before a question reaches the model. With zero retention on, nothing is stored.
- [Security and Privacy for Your AI Assistant](https://www.kelu.dev/features/security-privacy/): Kelu isolates every workspace, can mask PII before it reaches the model, never trains models on your data, and offers SSO, audit logs and permission controls.
- [Simulations: Test AI on Past Tickets First](https://www.kelu.dev/features/simulations/): Run the AI against your past resolved tickets before it answers live ones. Review each answer and its confidence, then turn on deflection step by step.
- [Slack Bot: Cited Answers Where Your Team Asks](https://www.kelu.dev/features/slack-bot/): Add the Kelu bot to Slack and get cited answers from your knowledge bases in channels, threads and DMs, instead of pinging the one person who knows.
- [Support Automation for Your Helpdesk](https://www.kelu.dev/features/support-automation/): AI for Zendesk, Intercom, Freshdesk, Front and Linear: deflect tickets, draft cited replies, triage, and test on past tickets before going live.
- [Support Form Deflector: Answer Before a Ticket](https://www.kelu.dev/features/support-form-agent/): Kelu sits in your support form and answers the question with cited sources before the ticket is sent, so only the hard cases reach your team.
- [Support Ticket AI: Answer Tickets, Draft Replies](https://www.kelu.dev/features/support-ticket-ai/): AI that answers tickets it has evidence for and tags the rest, inside Zendesk, Intercom, Freshdesk, Front and Linear. Draft-only mode available.
- [Microsoft Teams Bot: Internal Answers in Teams](https://www.kelu.dev/features/teams-bot/): Employees ask about policies, IT and products in Microsoft Teams and get cited answers from OneDrive, Confluence and the rest of your indexed knowledge.
- [Docs AI: AI Chat, Cmd+K Search and Sidebar](https://www.kelu.dev/features/website-widget/): Turn your documentation into an interactive AI assistant. Add Docs AI with one script tag: floating chat, Cmd+K search bar, or docked sidebar.

## Comparisons
- [AI Ticket Answers for Zendesk, Intercom & Freshdesk](https://www.kelu.dev/compare/ai-helpdesk-answers-zendesk-intercom-freshdesk/): Zendesk AI agents, Intercom Fin and Freddy AI vs helpdesk-agnostic tools like Kelu and eesel: where each works, what it reads, pricing, testing and citations.
- [Best AI Support Platforms for Technical Products](https://www.kelu.dev/compare/ai-support-platforms/): Kelu, Kapa, Inkeep, Intercom Fin, Zendesk AI and Chatbase compared on retrieval, helpdesks, chat surfaces, MCP, free tiers and pricing models.
- [Best Ways to Serve Your Docs Over MCP (2026)](https://www.kelu.dev/compare/best-mcp-servers-for-documentation/): How to give Cursor, Claude, VS Code and Windsurf access to your docs over MCP: Kelu, Kapa.ai, Inkeep, Mintlify, GitBook, Context7 and DeepWiki compared.
- [Build vs Buy an AI Docs Assistant: Real Costs (2026)](https://www.kelu.dev/compare/build-vs-buy-ai-docs-assistant/): What it costs to build your own docs AI on the OpenAI Responses API, LangChain or LlamaIndex vs buying one: time, engineers, connectors, ops and analytics.
- [Kapa.ai vs Inkeep vs Kelu: Three-Way Comparison](https://www.kelu.dev/compare/kapa-vs-inkeep-vs-kelu/): Kapa.ai vs Inkeep vs Kelu side by side: retrieval, source coverage, chat surfaces, MCP servers, helpdesk AI and pricing model in one matrix, checked Oct 2026.
- [Best RAG Platforms for Docs and Support (2026)](https://www.kelu.dev/compare/rag-platforms-for-docs/): How Kelu, Kapa, Inkeep, Chatbase, Brainfish and OpenAI File Search handle hybrid search, reranking, abstention, citations and MCP, from each vendor's docs.
- [Kelu Alternatives: Compare AI Knowledge Assistants](https://www.kelu.dev/alternatives/): See how Kelu compares to Kapa.ai, Inkeep, Glean, Intercom Fin, Zendesk AI, Guru and more on retrieval, sources, chat surfaces, API access and pricing.

## Guides (blog)
- [AI for Company Policies: From PDFs to Instant Answers](https://www.kelu.dev/blog/ai-for-company-policies/): Policies only work if people find them at the moment of decision. How retrieval-augmented AI turns a policy library into answers employees actually use.
- [How to Build an Internal Knowledge Assistant](https://www.kelu.dev/blog/build-internal-knowledge-assistant/): The five-layer anatomy of an internal AI assistant — connectors, indexing, retrieval, governance, and interfaces — and a build-vs-buy framework for each.
- [Company Brain: What It Means and How to Build One](https://www.kelu.dev/blog/company-brain-how-to-build/): Why wiki migrations fail, and how RAG builds a company brain across GitHub, Jira, Confluence, Notion and Drive, answering in Slack and Teams.
- [The Employee Handbook Nobody Reads, and How AI Helps](https://www.kelu.dev/blog/employee-handbook-ai-assistant/): Handbooks fail because they are optimized for legal completeness, not for answering questions. An AI assistant lets both goals coexist.
- [GEO for Documentation: Get Your Docs Cited by AI](https://www.kelu.dev/blog/geo-for-documentation/): Generative engine optimization is SEO's successor problem: your docs now need to win citations inside AI answers, not just rankings on a results page.
- [Rolling Out Internal AI Without Scaring Legal](https://www.kelu.dev/blog/governed-ai-for-internal-knowledge/): The fastest way to get an internal AI assistant blocked is to skip governance. Here is the checklist that gets policy and playbook Q&A approved.
- [How to Deflect Support Tickets Before Submission](https://www.kelu.dev/blog/how-to-deflect-support-tickets/): Auto-replies come too late: the ticket is already filed. Form deflection answers the question when the user presses submit.
- [How to Evaluate RAG: Recall, Grounding, Abstention](https://www.kelu.dev/blog/how-to-evaluate-rag/): Without a golden set and the right metrics, you can't tell a RAG improvement from a regression. The stack: retrieval metrics, answer metrics, abstention tests.
- [Hybrid Search Explained: Vector + Keyword with RRF](https://www.kelu.dev/blog/hybrid-search-explained/): Semantic search misses exact IDs; keyword search misses paraphrases. How hybrid search fuses both with reciprocal rank fusion, and why to start with it.
- [Best Kapa.ai Alternatives for AI Docs in 2026](https://www.kelu.dev/blog/kapa-ai-alternatives/): The top Kapa.ai alternatives for technical docs, developer portals and support automation in 2026: Kelu, Inkeep, Chatbase, Mintlify, GitBook and Glean.
- [Knowledge Gap Analysis: Build a Docs Roadmap](https://www.kelu.dev/blog/knowledge-gap-analysis/): A knowledge gap is a real question your content can't answer. Tracked systematically, gaps turn gut-feel docs planning into a ranked, evidence-based backlog.
- [MCP Server Guide: Your Docs in Cursor and Claude](https://www.kelu.dev/blog/mcp-server-for-docs/): The Model Context Protocol lets AI assistants query your live docs instead of guessing from training data. What an MCP docs server does and why it matters.
- [Permission-Aware AI Knowledge Base With RBAC](https://www.kelu.dev/blog/permission-aware-ai-knowledge-base/): Stop AI assistants leaking confidential docs across teams with document-level visibility, role-based access and scoped retrieval.
- [Playbooks That Answer Back: Making SOPs Operational](https://www.kelu.dev/blog/playbooks-that-answer-back/): A playbook nobody consults is documentation theater. Turning SOPs into an answer engine is how process actually changes behavior.
- [RAG vs Fine-Tuning for Docs Q&A: Which Do You Need?](https://www.kelu.dev/blog/rag-vs-fine-tuning/): For Q&A over your own docs, retrieval-augmented generation beats fine-tuning on freshness, citations, and cost. A decision framework for choosing between them.
- [AI Search for Your Docs Site: Mintlify, Docusaurus](https://www.kelu.dev/blog/replace-docs-site-search-with-ai/): Swap keyword docs search for search with cited AI answers. Setup on Mintlify and Docusaurus, what works on GitBook, and how to measure the change with gaps.
- [RAG for Runbooks: Faster Incident Response](https://www.kelu.dev/blog/runbooks-rag-incident-response/): During an incident, the cost of finding the runbook is paid in downtime. Retrieval turns your runbook library into an on-call copilot.
- [Sales Playbooks That Answer Questions Mid-Call](https://www.kelu.dev/blog/sales-playbooks-in-the-flow/): Enablement content fails at the moment of the objection. Putting the playbook behind retrieval puts the approved answer in the rep's hands mid-call.
- [Slack AI Knowledge Bot: Answers Where Work Happens](https://www.kelu.dev/blog/slack-ai-knowledge-bot/): Knowledge tools fail when they demand a context switch. Putting a grounded AI assistant inside Slack meets questions where they are already being asked.
- [Test an AI Support Agent on Past Tickets First](https://www.kelu.dev/blog/test-ai-support-agent-before-go-live/): Replay resolved tickets through the AI, run Draft only for a week, and set the confidence threshold from real answers before an AI support agent goes live.
- [What Is Ticket Deflection? Definition and Formula](https://www.kelu.dev/blog/what-is-ticket-deflection/): Ticket deflection is the share of support questions resolved without creating a ticket. Here is how to define it, measure it honestly, and raise it.
- [Why AI Chatbots Hallucinate: Citations and Abstention](https://www.kelu.dev/blog/why-ai-chatbots-hallucinate/): Hallucination is a property to engineer around, not a bug to patch. How grounding, citations, and enforced abstention make a chatbot a trustworthy answerer.

## Glossary
- [What Is Abstention in AI Answers?](https://www.kelu.dev/glossary/abstention/): When an AI assistant declines to answer because its sources don't cover the question, and says so plainly instead of writing a confident guess.
- [What Is Chunking in RAG?](https://www.kelu.dev/glossary/chunking/): Splitting documents into smaller passages before embedding, so retrieval can return the precise section that answers a question rather than a whole page.
- [What Are Citations in AI Answers?](https://www.kelu.dev/glossary/citations/): Links from an AI answer back to the specific source documents used to generate it, allowing users to verify every claim.
- [What Is Deflection Rate?](https://www.kelu.dev/glossary/deflection-rate/): The percentage of support questions resolved by self-service — typically an AI assistant — without creating a ticket or reaching a human agent.
- [What Are Embeddings in AI Search?](https://www.kelu.dev/glossary/embeddings/): Numeric vector representations of text in which semantically similar passages end up close together, enabling search by meaning rather than exact words.
- [What Is Support Form Deflection?](https://www.kelu.dev/glossary/form-deflection/): Answering a support request inside the contact form, before the ticket is sent, so customers with documented questions get help without waiting for an agent.
- [What Is Grounding in AI Answers?](https://www.kelu.dev/glossary/grounding/): Tying an AI answer to specific source content, so each claim comes from retrieved passages rather than the model's memory and can be checked against them.
- [What Is AI Hallucination?](https://www.kelu.dev/glossary/hallucination/): AI hallucination is when a language model writes fluent, confident text that is factually wrong or made up. Why it happens and how grounding and citations help.
- [What Is Hybrid Search in RAG?](https://www.kelu.dev/glossary/hybrid-search/): A retrieval strategy that combines semantic (vector) search with keyword (full-text) search and merges the results, capturing both meaning and exact matches.
- [What Is a Knowledge Base?](https://www.kelu.dev/glossary/knowledge-base/): A curated, indexed collection of an organization's knowledge — docs, wikis, tickets, repos — that an AI assistant retrieves from when answering questions.
- [What Is a Knowledge Gap (Coverage Gap)?](https://www.kelu.dev/glossary/knowledge-gap/): A question real users ask that your content can't answer, or answers badly. Tracked over time, knowledge gaps become a ranked list of docs to write next.
- [What Is LLM Evaluation (Evals)?](https://www.kelu.dev/glossary/llm-eval/): A systematic method for measuring the quality of AI-generated answers — often using a second model as a judge — so quality is tracked rather than assumed.
- [What Is an LLM Gateway?](https://www.kelu.dev/glossary/llm-gateway/): An abstraction layer between an application and language-model providers, allowing models to be swapped, routed, and managed without changing application code.
- [What Is llms.txt and Do AI Engines Use It?](https://www.kelu.dev/glossary/llms-txt/): A proposed standard for a markdown file at /llms.txt that gives AI tools a short summary of a site and links to its most useful pages, so they skip the clutter.
- [What Is MCP (Model Context Protocol)?](https://www.kelu.dev/glossary/mcp/): An open protocol that lets AI applications (like IDE assistants) connect to external tools and data sources through a standard JSON-RPC interface.
- [What Is PII Masking in AI Assistants?](https://www.kelu.dev/glossary/pii-masking/): Finding personal data such as emails, phone numbers and card numbers in text, and replacing it with placeholders before an AI model reads it or it is stored.
- [What Is Prompt Injection?](https://www.kelu.dev/glossary/prompt-injection/): An attack in which adversarial instructions are placed in user input or in retrieved content to manipulate an AI system into ignoring its rules or leaking data.
- [What Is RAG (Retrieval-Augmented Generation)?](https://www.kelu.dev/glossary/rag/): RAG retrieves relevant documents from a trusted source and hands them to a language model, so answers come from your real content, not the model's memory.
- [What Is Reciprocal Rank Fusion (RRF)?](https://www.kelu.dev/glossary/reciprocal-rank-fusion/): A way to merge ranked result lists into one by scoring each item on its position in each list, not its raw score, so different search methods can be combined.
- [What Is Reranking in RAG?](https://www.kelu.dev/glossary/reranking/): A second pass in which a stronger model re-scores the top search results against the question, so the passages that actually answer it rise to the top.
- [What Is Semantic Search?](https://www.kelu.dev/glossary/semantic-search/): Search that ranks results by meaning similarity between the query and documents, using embeddings, rather than by matching keywords.
- [What Is a Vector Database?](https://www.kelu.dev/glossary/vector-database/): A data store optimized for saving embedding vectors and finding the nearest neighbors of a query vector quickly, at scale.
- [What Is Zero Data Retention (ZDR)?](https://www.kelu.dev/glossary/zero-data-retention/): A setting under which an AI service answers a request and then discards it: the question and answer are not written to history, logs or traces.

## Optional
- [Integrations: Docs, Helpdesks, Chat and Code](https://www.kelu.dev/integrations/): Connect Kelu to docs sites, Confluence, Notion, Google Drive, GitHub, Zendesk, Jira, Slack, Teams and more. One answer pipeline for every place people ask.
- [AI Use Cases for Support, Docs, DevEx and More](https://www.kelu.dev/use-cases/): Explore 30+ ways teams use Kelu: ticket deflection, docs Q&A, developer portals, internal helpdesks, community bots, ticket copilots, and more.
- [Solutions by Team: AI Answers for Every Team](https://www.kelu.dev/solutions/): How support, DevRel, engineering, sales, HR, IT helpdesk, customer success, and product teams use Kelu for instant, cited AI answers from their own knowledge.
- [Industries — AI Knowledge Assistant by Industry](https://www.kelu.dev/industries/): How SaaS, fintech, healthcare, e-commerce, developer tools, education, gaming, telecom, manufacturing and public-sector teams use Kelu for cited AI answers.
- [Enterprise: SSO, Unlimited Usage and a Custom SLA](https://www.kelu.dev/enterprise/): Kelu Enterprise: SSO with OIDC or SAML 2.0, source groups, unlimited knowledge bases and questions, a white-label widget and a custom SLA.
- [Data Processing Agreement (DPA)](https://www.kelu.dev/dpa/): Kelu's Data Processing Agreement for GDPR: what personal data we process for customers, security measures, sub-processors, transfers and data subject rights.
- [Privacy Policy: How We Handle Your Data](https://www.kelu.dev/privacy/): How Kelu collects, uses, shares and protects your personal information and the content you connect, how long we keep it, and the rights you have over it.
- [Terms of Service and Acceptable Use](https://www.kelu.dev/terms/): The terms for using Kelu: accounts, acceptable use, subscriptions, payments and refunds, who owns your content, service availability, liability and termination.
- [Full text for LLMs](https://www.kelu.dev/llms-full.txt)