Context Manager

A brand knowledge base for AI is not a document repository. It is the structured, searchable set of facts, rules, and definitions that an AI agent must know to act as your representative. It includes your positioning, your tone, the claims you make and the ones you avoid, your key audiences, and your competitors.

This matters now because your buyers are asking AI engines like ChatGPT, Gemini, and Google AI Overviews what to buy. The answer they get is a single paragraph, not ten blue links. If the AI answering them does not know your brand's ground truth, it cannot represent you accurately. Your own AI marketing tools, if they write from generic models, will drift off message. A dedicated brand knowledge base anchors all AI-generated work to your reality.

Context Manager is the screen at /dashboard/knowledge-bases. It has three sections: Brand Context, Knowledge Base, and Audience Segments. The AI Marketer and every Agent you build read from these sections before they write or act.

Brand Context holds your core positioning, tone guidelines, and competitor list. This is your brand kit for AI content. When the AI Marketer drafts a landing page, it checks these rules first. Knowledge Base is where you upload or paste your source-of-truth material, like product specs, approved messaging, and FAQs. The AI can search this base and answer questions, citing the specific passage it used. Audience Segments let you define who you are writing for, like 'procurement lead at a mid-market SaaS company' or 'founder of a seed-stage startup'. This detail ensures a draft for one segment reads differently from a draft for another.

The system does not guess or infer. It reads exactly what you provide. There is no automatic distillation from call transcripts or Slack. You add and maintain the facts. When you update a positioning statement, the AI Marketer uses the new version in its next task.

Here are two concrete workflows. First, briefing an AI Agent for a new campaign. Open Context Manager and update the Brand Context with the campaign's key message and any competitor mentions to avoid. Then, in the Knowledge Base, add the campaign brief and target persona details. Finally, ask the AI Marketer in chat to draft the campaign's launch email. It will pull from the updated context, ensuring the draft is on message.

Second, scaling personalized outreach. Define your Audience Segments in detail, such as 'CTO at a company using legacy monitoring tools'. In Opportunities, filter for journalist requests matching that expertise. Ask the AI Marketer to draft a pitch. It will use the relevant audience segment to tailor the language, making it specific to a technical leader's concerns. You review and send the approved draft from your own mailbox.

Ground truth is the non-negotiable facts your AI must never hallucinate. In Context Manager, this is your Knowledge Base. When you ask the AI Marketer a question like 'What are our three key differentiators?', it searches your uploaded documents and returns an answer that cites the paragraph where you stated them. This citation is shown in the chat.

This process creates a clear audit trail. You can see which source the AI used for any claim. It also limits drift. Without this anchor, an AI agent might gradually incorporate external, off-brand information from its base model. With Context Manager, every generation is constrained and directed by your materials first.

Context Manager has specific limits. Automatic distillation of facts from call transcripts, Slack channels, or email threads is not built. You must add the facts manually. It does not proactively scan your connected apps to build a knowledge base.

It is also not the chat. Images or video clips generated by the AI Marketer are kept within the chat or document where they were created. Context Manager only handles text-based brand knowledge, audience definitions, and rules.

Context Manager is live on every Mentionry plan. The free seven day trial includes it, so you can set up your brand kit and test the AI Marketer with your own context. The Starter plan at $99 a month includes Context Manager with one brand and one seat. The quoted Enterprise plan includes it for all the customer's brands under one login, with more seats and API access.

Recording begins when you add your first prompt and context. There is no backfill of historical data. If your brand is never mentioned in answer engines, you will not have an average position metric; you start from zero.