--- title: Client Brain — AI-Powered Knowledge Base Initiative type: article created: '2026-04-05' updated: '2026-04-05' source_docs: - raw/2026-03-12-papertube-sync-129326591.md tags: - ai-tools - process-improvement - client-brain - ai-content-governance - knowledge-management layer: 2 client_source: null industry_context: null transferable: true --- # Client Brain — AI-Powered Knowledge Base Initiative ## Overview A recurring problem across client accounts is that institutional knowledge about client business models, terminology preferences, and content rules lives in people's heads rather than in a shared, queryable system. When this knowledge isn't captured, content errors slip through — wrong business model descriptions, forbidden phrases, or factual misrepresentations — that erode client trust and create rework. This initiative proposes building a structured **Client Brain knowledge base** for every account, using AI to actively govern content against client-specific rules before it goes out the door. The concept surfaced during a pre-call prep session for the [[wiki/clients/bluepoint/index|Bluepoint]] account, where a social post incorrectly described reverse ATMs as leased (they are owned by Bluepoint; business owners benefit from foot traffic, not revenue). Neither the account manager nor the strategist knew the correct model. A similar pattern exists with [[wiki/clients/citrus-america/index|Citrus America]], where terms like "peel oil," "crushing," and "chocolate" (vs. "chocolate flavored") have caused repeated friction. --- ## The Problem Client-specific knowledge accumulates informally over the course of an engagement but is rarely written down in a structured way: - **Business model nuances** — who owns what, who pays whom, how value flows to each party - **Forbidden terminology** — words, phrases, or framings the client has explicitly rejected - **Competitive sensitivities** — topics or comparisons to avoid - **Audience insights** — what resonates on sales calls, what objections come up Without a system to capture and enforce these rules, the same mistakes recur. Content writers, designers, and strategists onboarding to an account have no single source of truth. > *"We're building institutional knowledge about these clients, but we're not doing a particularly good job of writing it down anywhere."* > — Mark Hope --- ## Proposed Solution ### 1. Client Brain Knowledge Base Section Add a dedicated **Institutional Knowledge** section to each client's existing Client Brain workspace. This section would contain: - Business model summary (who owns what, revenue flows, customer value props) - Terminology rules (approved language, forbidden phrases, preferred framings) - Content guardrails (topics to avoid, claims that require verification) - Sales intelligence (common objections, what resonates with prospects) - Running learnings log (updated after each client call) ### 2. AI Content Governance Layer Feed the knowledge base into an AI content-checking workflow. Before any content (blog post, social copy, ad creative, web page) is sent to the client, it is run against the client's rules: - Flag forbidden terms or phrases - Catch factual claims that contradict the documented business model - Surface missing context that the client has previously flagged This replaces the current manual checklist approach (which exists for Bluepoint but is not systematized across accounts) with an AI-assisted pass that scales across all clients. **Example prompt pattern:** > *"Review this content against the following client rules: [rules]. Flag any violations or potential issues."* ### 3. 30-Minute Discovery Call Per Client To seed the knowledge base, schedule a focused 30-minute discovery call with each client using a standard question set: - What is your business model? Where does revenue come from? - Who are your primary customers and what do they care about? - What terminology do you prefer or want to avoid? - What do prospects push back on? What resonates? - What have we gotten wrong in the past that we should never repeat? These calls are framed as a quality initiative, not a remediation exercise. The output feeds directly into the Client Brain knowledge base. **Suggested question areas:** | Category | Example Questions | |---|---| | Business model | Who owns the product/equipment? Who pays whom? | | Terminology | Any words or phrases we should avoid? Preferred alternatives? | | Audience | What do leads care about most? What objections come up? | | Competitors | Who should we not mention or compare against? | | Past misses | What content have we produced that wasn't right? | --- ## Implementation ### Immediate Next Steps - [ ] **Mark Hope** — Schedule a call with Evoke and Gavin to discuss AI content governance workflow and how to implement the Client Brain knowledge section *(action item from 2026-03-12 prep call)* - [ ] Develop standard discovery call question template - [ ] Pilot with Bluepoint (given active friction) and Citrus America (given recurring terminology issues) - [ ] Roll out 30-minute discovery calls to remaining accounts within 30 days ### Ongoing Process - At the end of each client call, capture learnings: *"What did we hear or see this week that we should add to the knowledge base?"* - Assign ownership of each client's knowledge base to the account manager - Review and update rules whenever a client flags a content error --- ## Known Client Examples | Client | Known Rule / Learning | |---|---| | [[wiki/clients/bluepoint/index\|Bluepoint]] | Reverse ATMs are **owned** by Bluepoint, not leased. Business owners benefit from foot traffic only, not revenue share. | | [[wiki/clients/citrus-america/index\|Citrus America]] | Do not mention "peel oil." Use "squeezing," not "crushing." Flavoring must be described as "chocolate flavored," not "chocolate." | --- ## Related - [[wiki/clients/bluepoint/index|Bluepoint — Client Index]] - [[wiki/meetings/2026-03-12-bluepoint-pre-call-prep|Bluepoint Partnership Strategy Call — Pre-Call Prep]] - [[wiki/knowledge/account-management/client-reporting-standards|Client Reporting Standards]]