Website pages and PDFs
Import public website pages and text-based PDFs. Supported JavaScript-rendered pages can be crawled too. Inspect the extracted text and import failures before relying on a source.
AI chatbot trained on your website
Your website becomes the chatbot’s source of truth.
Obseri crawls the pages you already maintain and turns them into a searchable, inspectable knowledge base. The resulting assistant answers from that evidence instead of hiding everything behind vague AI training.
Visible pages, blocks, and citations
Incremental refresh when content changes
Retrieval testing before launch
Know exactly what the chatbot learned—and what it did not.
Obseri exposes discovered pages, extracted content, revisions, crawl activity, errors, and the knowledge blocks used during retrieval.
Clean, traceable knowledge
Semantic search retrieves related content even when visitors use different wording, with keyword fallback when needed. Inspect the supporting sources and test important answers before publishing.
Fresh without relearning
Conditional requests, content hashes, and revision history identify real changes so a refresh can preserve stable knowledge and update only what moved.
How it works
Crawl the public site
Start with one URL. Obseri checks crawl rules, discovers important pages, and shows each stage instead of leaving you at an unexplained loading screen.
Inspect and test retrieval
Search learned pages, open extracted content, add manual knowledge, and ask test questions to see which sources the assistant would use.
Publish with guardrails
Define unknown-answer behavior, escalation, personality, and voice, then embed the tested assistant with a small asynchronous script.
Built for questions scattered across real websites.
SaaS product and pricing
Answer questions across feature pages, pricing, documentation, integrations, trust content, onboarding material, and release notes.
Documentation search
Turn long technical guides into conversational retrieval while preserving direct links to the page that supports each answer.
Service businesses
Explain services, availability, process, policies, and next steps without forcing visitors to understand the site structure first.
Frequently changing content
Refresh product information, schedules, policies, and documentation while retaining revision history and visibility into what changed.
Clear answers
What buyers ask.
How does Obseri train a chatbot on my website?
Obseri does retrieval-based knowledge ingestion rather than retraining a foundation model. It discovers pages, extracts useful content, creates searchable blocks, and retrieves relevant evidence for each visitor question.
Can I see the content it collected?
Yes. Soul Studio exposes the source, individual documents, extracted content, block counts, crawl status, revisions, and retrieval results so you can audit the knowledge base.
What happens when my website changes?
Refresh uses validators and content hashes to recognize unchanged pages and create revisions for meaningful updates. You can inspect the resulting crawl activity and page state.
Can the chatbot invent answers outside my content?
You can require grounded behavior and set instructions for how the agent responds when retrieval does not contain enough evidence. The test lab lets you evaluate this before publishing.
Let visitors talk to your website.
Give Obseri one public URL. Review what it learns, shape the agent, and test the real visitor experience before you publish.
Start with your website