We’re excited to introduce the Knowledge Base for CometChat’s Agent Builder. This new capability gives your AI agents a centralized, governed source of truth by letting you collect, index, and attach knowledge sources directly from the CometChat Dashboard.
With Knowledge Base, every document, note, or link your agent relies on lives in one shared workspace. Content is indexed, searchable, and attachable per agent, ensuring responses stay aligned with your domain expertise, internal policies, and historical context, without duplicating data or managing scattered files.
Why does this matter to me, as a developer?
As you build AI-powered chat experiences, the quality of your agent’s responses depends heavily on the quality and consistency of its underlying knowledge. Without a structured way to manage that knowledge, agents can become hard to maintain, inconsistent across use cases, or prone to serving outdated information.
Knowledge Base helps you:
Centralize all agent knowledge in one governed workspace instead of managing ad hoc uploads per agent.
Reuse the same sources across multiple agents while keeping each agent’s context intentionally scoped.
Quickly update, detach, or retire knowledge without redeploying or reconfiguring agents.
The result is faster iteration, cleaner agent configurations, and more reliable responses in production.
Key features and capabilities
Centralized source management
Upload and curate PDFs, DOCX, TXT files, inline text entries, and web links in a single Knowledge Base. Every source is indexed and searchable, acting as shared memory for your agents.
Granular source attachment
Attach or detach individual sources per agent with a toggle. There’s no need to re-upload content—one source can power multiple agents while remaining centrally managed.
Multiple ingestion methods
Files: Batch upload up to 10 documents at a time.
Text: Create lightweight notes or SOPs directly in the dashboard.
Links: Index individual URLs or scrape and selectively index linked pages from a website.
Integrations: External integrations are coming soon to keep knowledge continuously in sync.
Status tracking and reliability
Each source clearly shows whether it is Indexed, Indexing, or Failed. This makes it easy to catch issues early and ensure agents only pull from healthy, processed data.
Safe updates without data loss
Detaching a source immediately removes it from an agent’s retrieval context without deleting it from the Knowledge Base, making updates reversible and low risk.
Interesting tech details
Under the hood, Knowledge Base chunks and indexes every source so agents can retrieve specific, relevant passages instead of relying on entire documents. This improves both response accuracy and citation clarity during testing and production conversations.
The attachment model is intentionally decoupled from ingestion. Index once, attach many times. This architecture keeps shared knowledge consistent across agents while still allowing precise control over what each agent can access.
Combined with indexing status visibility, this design supports scalable agent deployments without sacrificing governance or performance.
Reference links
Knowledge Base gives you a structured, scalable way to manage what your AI agents know, so you can focus on building smarter, more reliable conversational experiences.
Nivedita Bharathy
Product Marketing Specialist , CometChat