Loads your world in — and answers with sources.
Nine formats, one index. zep answers every question exclusively from the material you upload — and names the source for every sentence it pulled it from. Nothing made up, nothing hallucinated, no third-party training data.
Six source types, one unified index.
No matter which tool your knowledge comes from — we unify, split and search it. You don't have to structure anything manually.
Multi-page documents, tables, OCR on scanned images. Chapter structure and page numbers are preserved for source citations.
Notion
Workspace via OAuth, select individual databases or pages. Re-sync automatically every hour or trigger manually.
Websites / URLs
Sitemap crawl, parse main content, ignore footer / cookie banner / navigation. robots.txt is respected.
FAQ lists
Question-answer pairs as CSV or JSON. Indexed with higher priority — perfect answers for the most common questions.
Discord exports
Export a Discord server as JSON, upload it — the bot learns from community discussions without needing an active bot user on the server.
Markdown / TXT / JSON / ZIP
Plaintext, Markdown with frontmatter, structured JSON, ZIP archives with mixed formats. Everything is unified and chunked.
Example · “How long does your delivery to Austria take?”
Standard shipping to Austria takes 3–4 business days from shipping confirmation. An express option has been available since January 2026 — the parcel then reaches you on the second business day, for a surcharge of €8.90. Both options are shown at checkout.
Three knowledge problems classic bots don't solve.
- Hallucinations. Classic LLM bots invent answers from their training material when they're unsure. In consulting, support or legal contexts that's damaging to business.
- A PDF is not an API. PDF, Excel, Notion, website — all structured differently. A chatbot has to unify these sources, split them and make them searchable.
- Knowledge goes stale. You change a pricing page, add an FAQ, add a shipping note — and the bot knows nothing about it and answers with last month's data.
Six tasks we handle for you.
Vector search, re-ranking, tenant isolation, GDPR — sounds like an engineering sprint. At Zeptix it's the default; you never see these points as a task.
Vector search with re-ranking
First stage: embedding search over pgvector. Second stage: BM25 + cross-encoder re-ranking. Answer relevance clearly beats simple cosine similarity.
Strict tenant isolation
Your data lands in your workspace, never cross-indexed with other customers. Embeddings are tenant-scoped, no leakage between bots.
Re-indexing in the background
You upload new data — the bot stays responsive with the previous version until re-indexing finishes. No downtime, no progress bar.
Tagging + filtering per answer
You can tag sources ("internal", "public", "archived") and configure in the bot setup which tags are used per answer.
GDPR-compliant hosting
Embeddings + original data are stored in the EU. The DPA is available for download in the workspace; deleting sources takes effect immediately across all indexes.
Marketplace as a knowledge pack
No material of your own? Activate a pack with curated knowledge — maintained by the Zeptix editorial team, updated monthly.
FAQ on the knowledge base.
How exactly does RAG (Retrieval-Augmented Generation) work?
Before the LLM answers, we search your knowledge base for the passages most relevant to the question. These passages are sent to the LLM together with the question — and it answers exclusively based on that context. As a result, hallucinations statistically occur far less often.
Is my data used to train third-party models?
No. We use LLM providers (e.g. OpenAI, Anthropic, Mistral) exclusively in inference mode with an explicit opt-out for training data. Your embeddings and original sources stay in our EU database.
How large can my knowledge base be?
On the Free plan: 100 MB per bot. Plus: 5 GB. Pro: 50 GB. Size limits are soft — we contact you when you approach the limit instead of simply cutting you off.
Can I protect sources so only logged-in users see them?
Yes. For each source you can set a visibility level (public, login, role-restricted). Public bot answers draw only from public sources, while logged-in users get extended context.
How reliable are the source citations?
Very high. For every statement the bot shows the ranked top-3 matches from the vector search. If the answer is not backed by those sources, it flags it as "not directly supported by sources".
What happens when I delete a source?
It disappears immediately from the vector index, embeddings are deleted, and the bot can no longer use it for answers. The audit log keeps the deletion entry for 90 days as proof.
Pack in your knowledge. zep does the rest.
One PDF, one Notion workspace, one website crawl is enough — the bot is ready instantly and answers with source citations on the very first test.
