Source of Truth

One article. Customer, agent, and AI all read the same text

SOPs scattered across personal notes drift apart within a week. volbor keeps documentation in one place: the portal, inbox sidebar, and agent search index. The agent cites the article — it's not 'always right'.

Portal

Customers self-serve while your team handles escalations

Categories, typo-tolerant search, mobile responsive layouts, tables, and video. Custom subdomain and brand colors — no third-party branding in the header. SEO includes sitemaps and meta tags, not 'guaranteed organic traffic'.

Public & Restricted Access

Set your portal to public, password-protected, or SSO-only. Internal operational SOPs shouldn't show up in search.

On-Site Widget

Search articles directly in the header or chat widget: users don't have to leave for another browser tab unless you configure it that way.

In-Messenger Cards

Bots send concise article snippets. A 'Talk to human' button is always available if the text doesn't resolve the question.

Languages

Localized article versions. Machine translation acts as a starting draft for editors, not an unsupervised publish.

Sidebar

Find the snippet — insert into the reply. No Google Drive tabs needed

Operators search right in the chat panel and copy approved excerpts in one click. Internal guidelines are visible by role permissions. Timestamps and authors ensure no one sends last year's pricing.

RAG

AI agents answer using articles and cite the source. Not 'absolute truth'

Your knowledge index is chunked into semantic fragments. If no relevant fragment is found, the agent won't hallucinate SKUs: it escalates to a human.[1] Update the price in an article, and the next answer reflects the change after re-indexing — not 'instantly across the entire universe'.

Direct Citations

Every reply references an exact paragraph and can attach the Help Center URL.

Knowledge Gaps

If returns for a premium line aren't documented, the agent gives an honest 'I'll transfer you to a specialist', not a made-up policy.

Instant Indexing

Chunking and indexing without code releases. Editors update text without asking engineering for new embeddings.

Editorial Workflow

Drafts, approvals, rollbacks. No stealth edits directly in production

Full version history, inline comments, and section owners. AI and your portal read only published revisions, never someone's unapproved discount draft.

  1. v18Mara · ShippingPublished
  2. v19Leo · DraftIn Review
  3. v17RollbackIf v18 broke the pricing table

Knowledge Gaps

Search queries with zero hits are your content backlog, not 'bad customers'

Track zero-result searches, 'article didn't help' feedback, and which guides deflect tier-1 tickets.[2] Automated reminders for authors to review pricing after each release. Cohorts like 'questions after mobile app update' provide actionable signals for product teams, not reports for the sake of reports.

sofa delivery124
assembly / elevator41
premium return0

FAQ

Imports, languages, search engines

Only for existing customers?

Public, password, or SSO. Internal SOPs are protected with noindex and role-based permissions.

Import options?

Markdown, HTML, plain text files. Notion works via export, not a 'magical connector that never breaks'.

Can operators suggest edits?

Yes: create a draft or leave an inline comment directly from a ticket. The section editor reviews and publishes.

How fast does Google index articles?

Sitemaps update automatically. Crawling takes days, not minutes. That depends on search engines, not a volbor SLA.

Export with video embeds?

Full archive export including media assets and PDFs in a clean folder structure.

Auto-translation into 50 languages?

Generates a localized draft. Publishing happens after catalog terminology is verified by a human editor.

Getting Started

Consolidate pricing into a single article before training an AI agent

Launch your Help Center in volbor, embed snippets into the inbox, and enable RAG with source citations. Don't promise customers 'error-free AI'.

Open Workspace

Methodology & Notes

Where the figures on this page come from

  1. [1]
    No fragment in index — the agent won't hallucinate SKUs

    RAG behavior in the platform: answers are grounded strictly in chunked articles, otherwise triggering human escalation. This is a source constraint mechanism, not an empirical '100% accuracy' guarantee or hallucination benchmark.

  2. [2]
    Zero-result searches are your content backlog

    Help Center analytics report: tracked search queries and hit counts. The 'premium return → 0' row on this page is an illustrative example of a knowledge gap, not aggregate volbor customer statistics.