Skip to content
Knowledge base

Help articles your customers, your team, and your chatbot all read from

Write the answer once. The public part becomes a help center, the private part trains the AI, and the inbox drafts and the chat widget cite it back to you. Stop answering the same question in three places.

CommsOperator's knowledge base hosts a public help center per brand, with categories, articles, feedback boards, a roadmap, and a changelog, prerendered so every page loads fast and shows up in search engines. Internal training books hold the notes the AI may use but never shows to customers. Search combines meaning and keywords. The chatbot answers from your content with citations and hands off to a person when it should. SharePoint and OneDrive connectors are not available in the hosted service.

Capabilities

What changes for your team

Public help centers

Each brand gets its own help center, with its own branding, under your workspace path. Pages are prerendered, so they load fast and Google can index every article. Categories, slugs, and visibility controls decide what the public sees.

Training books

Give the AI the context you would give a new hire, without publishing it. Hierarchical notes, folders, and documents ground replies for specific channels and are never shown to customers.

Hybrid search

Finds the article even when the wording differs. Semantic search on OpenAI embeddings and keyword matching run together over your training notes, and one search box covers articles, notes, tasks, feedback, roadmap, and changelog entries.

Import and content tools

Move a whole document set in one upload. Bulk ZIP import brings training material in at once with AI tag suggestions on request, PDF and Word files are text-extracted, and your engineers' docs become help articles through Bitbucket sync without copy-paste. AI tools draft, improve, or summarise on request.

Grounded chatbot

Every answer shows which article it came from. The chatbot remembers the session, can capture an email, and escalates to the inbox with the full transcript when it should stop and fetch a person.

Knowledge gap detection

Know which questions your help center cannot answer yet. Unanswered questions from chat and the inbox surface as gaps, each linked to the closest existing articles so you know what to write.

Proof

Numbers from the production instance and behaviour checked against the code. No projections, no testimonials.

  • Help centers, feedback boards, roadmaps, and changelogs are live on 14 Tech Family Ventures websites
  • Every public article is prerendered, so search engines index it without running the application
  • Your other tools can read the help center too: an external knowledge base API with scoped project tokens allows headless use

Limits, stated plainly

What this module does not do today, and what to do about it. If one of these is a blocker, you know before you commit, not after.

  • SharePoint and OneDrive document connectors are not available in the hosted service. Bring documents in by ZIP, PDF, or Word upload, or through Bitbucket sync.
  • The chatbot answers only from your content. When nothing relevant exists it says so and offers a person, so an empty help center means a chatbot that mostly hands off.
Connects with

Works with what you already run

Bitbucket (docs sync)
OpenAI embeddings
Website chat widget
External KB API

Frequently asked questions

Can each brand have a separate help center?

Yes. Every workspace publishes its own portal with its own categories, articles, branding, and widgets. A member of one brand's workspace cannot see another brand's drafts.

Does the chatbot make things up?

It answers from retrieved articles and notes with citations. When nothing relevant is found, it says so and offers a handoff to a person instead of inventing an answer.

Can the chatbot answer without a person?

Yes, on your website, from your own articles and notes, with the source cited on every answer. When it finds nothing relevant, or the visitor asks for a person, it hands the conversation to the inbox with the transcript attached.

See knowledge base on your own data

Tell us what you run and we will be straight about the fit. Private beta. A person reads every request.