Skip to main content

Context and knowledge records

No. Profound prepopulates knowledge records from the data you’ve already given it and from background collection about your brand, so files appear even before you upload anything. Adding documents to your Knowledge Base and connecting synced sources deepens the context further.
Context Manager fills its knowledge records from the Brand Context, Knowledge Bases and synced data. Add each of those sources to the Context Manager:
  • Brand Context: select Add context to upload files that describe your brand as a whole.
  • Knowledge Base: add documents and webpages you want Profound to search as source material. See Knowledge Bases and Brand Kits.
  • Synced Data: connect Slack, Email, or Gong so Profound learns from conversations with your customers.
For the step-by-step setup, see Get started with Context Manager.
Brand guidelines, product information, positioning, approved claims, and editorial rules. For a longer list, see What kinds of content should I add to my Knowledge Base.
Yes. Knowledge records capture voice, approved claims, and do-and-don’t rules, and AI Marketer and Agents draw on them when generating content. Correct a section once, and the change applies wherever your context is retrieved.
Knowledge Base uploads populate into knowledge records in about 5 minutes, and context from a Gong call appears about an hour after the call. Edits you approve through AI Marketer propagate almost immediately.
Open the section and fix the claim manually, or tell AI Marketer what to change. The fix applies everywhere that context is used.
Context ingestion runs continuously: new Knowledge Base documents and synced source activity keep the files fresh. When something is missing, Profound prompts you to fill the gaps through Identified gap cards in the Brand Context tab.
To opt out of ingesting context from a source, select the source in the Synced Data table and toggle off the Context ingestion setting.Only organization admins can change these settings. Learn more in Roles and permissions.
Audience Segments is a legacy feature that is now a part of the Brand Context. To define or edit your audience, go to Context Manager > Brand Context and refine your Audience knowledge record.

Knowledge Bases

A Knowledge Base contains the source material you give Profound: the documents, webpages, and files themselves. A knowledge record is what Context Manager distills from that material, organized into your company, audience, competitors, offerings, strategy, and execution.
Documents that describe how your brand works give the best results: brand guidelines, legal and compliance rules, product information, marketing strategies, content calendars, prior reports, sample outputs, and editorial standards. Anything that answers your priorities, your audience, your voice, and what to always do or avoid.See What makes a good Knowledge Base for FactCheck for useful best practices for Knowledge Base content. What works for FactCheck also works well for Context Manager.
Supported file types are: PDF, text-based formats (Markdown, plain text), CSV, Word documents and Excel spreadsheets. Profound parses all uploaded files to raw text before indexing, so non-text-based files may not display optimally. This doesn’t affect search.
An auto-sync feature is available for sitemap or URL crawls and external imports (like Notion and Google Drive). It’s off (set to Never sync) by default. Enable it in the import dialog or in the folder settings on the Knowledge Base page.
Enroll your domain in organization Settings > Web Scrape, then import the site with Crawl Website > Sitemap or add pages with Scrape Page. For an enrolled domain these options fetch pages with ProfoundBot, which you can allow by User-Agent or IP address. The Profound Knowledge Base Fetcher reads the sitemap file itself and has its own User-Agent.The Crawl option uses an external crawler that doesn’t identify itself as Profound, so it can’t be allowed by User-Agent or IP address. See Crawl Website for details.
Files uploaded to Knowledge Bases can’t exceed 100 MB. Split larger files into smaller ones and upload them separately.Increasing this limit is on our roadmap. Given how Profound processes information in Knowledge Bases (separating data into smaller chunks to enable effective search), it’s always best practice to keep files reasonably sized.