Purpose: This is your guidebook for building the Arcus Compliance and Libra knowledge base. This knowledge base becomes the single source of truth that powers every piece of communication: emails, cold outreach, landing pages, LinkedIn copy, proposals, and any message sent on behalf of the business.

This document explains what the knowledge base is, how it gets built, and how to use it once it exists.

What Is the Knowledge Base?

The knowledge base is a structured reference document that any AI (Claude, or any future tool) can read before writing anything for you. Think of it as briefing a new copywriter on day one. Instead of explaining Arcus from scratch every time, the AI reads this document and already knows your voice, your positioning, your clients, your regulations, and your boundaries.

It covers two brands:

Arcus Compliance -- your compliance consultancy helping regulated product brands navigate UK and EU regulatory requirements.

Libra -- (to be defined based on ChatGPT extraction. If Libra is a separate entity, product, or brand extension, the knowledge base will have its own dedicated section.)

How the Knowledge Base Gets Built

This is a two-phase process.

Phase 1: ChatGPT Extraction (You Do This)

You have already completed or are working through the ChatGPT Migration Worksheet. The mega-prompt and follow-up prompts will produce raw outputs across 15 sections. That raw material is the input.

Specifically, the sections that feed directly into the knowledge base are:

Section 2 (Business Overview) feeds into Brand Identity and Positioning

Section 3 (Compliance and Regulatory Knowledge) feeds into Domain Expertise

Section 4 (Clients and Accounts) feeds into Audience and ICP

Section 5 (Content and Marketing) feeds into Voice, Tone, and Content Rules

Section 6 (Sales and Outreach) feeds into Messaging Frameworks

Section 8 (Documents and Templates) feeds into Templates and Formats

Section 11 (IP and Original Frameworks) feeds into Proprietary Methodology

Phase 2: Claude Generates the Knowledge Base (Nandesh Does This)

Once the ChatGPT outputs land, Nandesh feeds them into Claude along with everything already known about Arcus from the existing Notion workspace, the content calendar, the ICP research, and the outreach workflows. Claude then writes the knowledge base into the template below.

The validation test: if the knowledge base Claude produces matches what you know to be true about your business, the ChatGPT extraction worked. If there are gaps or inaccuracies, we go back and run targeted follow-ups.

The Knowledge Base Template

Below is the template structure. Each section will be populated after the ChatGPT export is processed. For now, this serves as the skeleton and as your reference for what "complete" looks like.

KB-1: Brand Identity

What goes here: The foundational facts about each brand that never change conversation to conversation.

Arcus Compliance

Company name and legal entity

Founded (year)

Founder and key team members with roles

Headquarters and operating locations

Company mission (one sentence)

Company vision (one sentence)

What Arcus does (two to three sentence elevator pitch)

Core services offered (list each with a one-line description)

Industries served

Geographies served (UK, EU, international scope)

Libra

Entity name and relationship to Arcus

Purpose and positioning

Services or products offered

Target audience

How Libra differs from Arcus in market positioning

KB-2: Ideal Customer Profile (ICP)

What goes here: Detailed description of who Arcus sells to. This drives all outreach and content targeting.

Primary ICP

Job titles targeted

Company size (revenue, employee count)

Industry verticals

Geographic focus

Regulatory triggers (what makes them need Arcus right now)

Common pain points (in their own language)

Where they spend time online

How they currently handle compliance (in-house, other consultants, ignoring it)

Secondary ICP

Who else buys but is not the primary target

How the messaging shifts for this group

Anti-ICP (Who to Avoid)

Company types that are not a fit

Red flags in qualification

Common time-wasters

KB-3: Voice and Tone Guidelines

What goes here: How Arcus sounds across all channels. This is what makes copy feel like Lee wrote it, not a robot.

Core Voice Attributes

(e.g., Authoritative but approachable, Direct, No jargon for the sake of jargon)

Words and phrases Lee uses naturally

Words and phrases to never use

Sentence length preference

Paragraph length preference

How Lee opens emails / messages

How Lee closes emails / messages

Humor: when it is appropriate and when it is not

Formality spectrum by channel (LinkedIn vs email vs cold outreach vs website)

Tone by Context

Cold outreach: (describe the tone)

Follow-up messages: (describe the tone)

Client emails: (describe the tone)

LinkedIn posts: (describe the tone)

Website / landing page: (describe the tone)

Proposals and formal documents: (describe the tone)

KB-4: Messaging Frameworks

What goes here: The core arguments, value propositions, and persuasion structures that power all communications.

Primary Value Proposition

One-line value prop

Expanded value prop (three to four sentences)

The "so what" test: why should the prospect care right now

Pain Points and Agitation

Top 5 pain points the ICP feels (stated in their language)

Consequences of not solving these problems

Emotional triggers that drive action

Regulatory deadlines or enforcement actions that create urgency

Differentiation

How Arcus is different from competitors

What competitors get wrong

Why Arcus, why now

Proof points (case studies, numbers, results)

Objection Handling

"We already have compliance covered" -- response framework

"Too expensive" -- response framework

"Not a priority right now" -- response framework

"We will handle it in-house" -- response framework

(Add any others surfaced from ChatGPT extraction)

KB-5: Domain Expertise (Compliance and Regulatory)

What goes here: The technical knowledge that establishes credibility. This is what separates Arcus content from generic marketing.

Regulations Covered

(List each regulation with a one-line plain-English summary)

UK-specific regulations

EU-specific regulations (REACH, GPSR, etc.)

International / cross-border considerations

Regulatory Bodies

(List each body and what they enforce)

Key Compliance Concepts

Terms and definitions Lee uses in content and client work

Common misconceptions Lee corrects

Recent or upcoming regulatory changes

Arcus Methodology

How Arcus approaches a compliance engagement

Phases or steps in the process

What deliverables clients receive

Timeline expectations

KB-6: Content Pillars and Themes

What goes here: The recurring topics and angles that structure all content creation.

Content Pillars

Pillar 1: (name and description)

Pillar 2: (name and description)

Pillar 3: (name and description)

Pillar 4: (name and description)

(Add as many as surfaced from ChatGPT)

Recurring Series or Formats

(Any named series Lee runs on LinkedIn or elsewhere)

Post formats that perform well

Content types by funnel stage (awareness, consideration, decision)

Content Rules

What to always include in posts

What to never include

Hashtag strategy

Call-to-action patterns

Linking and tagging conventions

KB-7: Outreach and Sales Messaging

What goes here: Ready-to-use structures for cold email, LinkedIn outreach, follow-ups, and sales conversations.

Cold Email Framework

Subject line patterns that work

Opening line approach

Body structure

CTA style

Follow-up sequence (number of touches, timing, tone shift per touch)

LinkedIn Outreach Framework

Connection request message template

First message after connection

Follow-up sequence

Character limits and formatting rules (295-character Expandi splits)

Email Templates

Introduction email

Follow-up after no response

Follow-up after meeting

Proposal send email

Check-in with existing client

Re-engagement of cold lead

KB-8: Proof and Social Proof

What goes here: Evidence that backs up claims. Every piece of copy is stronger with proof.

Case Studies

Client name (or anonymized), problem, solution, result

Quotable metrics or outcomes

Testimonials

Direct quotes from clients

Context for each quote

Credentials and Authority Markers

Certifications, memberships, affiliations

Speaking engagements or media appearances

Publications or thought leadership pieces

Years of experience and notable career milestones

KB-9: Proprietary Frameworks and IP

What goes here: Original thinking that belongs to Arcus. These are the things that make Arcus methodology distinct.

Framework name and description

When and how to reference it in content

Visual representation (if one exists)

How it maps to the client journey

KB-10: Channel-Specific Rules

What goes here: Formatting and style rules that differ by channel.

Post length targets

Hook patterns

Line break and spacing conventions

Emoji usage (yes/no, which ones)

Image and carousel rules

Email (Warm)

Subject line length

Salutation style

Signature format

Attachment conventions

Cold Email

Deliverability considerations

Plain text vs HTML

Link usage rules

Personalization variables

Website / Landing Pages

Headline structure

Section ordering

CTA button copy conventions

Trust signals placement

Mobile considerations

KB-11: Boundaries and Guardrails

What goes here: Things the AI should never do or say when writing for Arcus.

Claims never to make (legal, financial, guaranteed outcomes)

Competitors never to name directly

Topics to avoid

Regulatory advice disclaimers required

Confidential information handling

Client names that should not appear in public content

Formatting rules that must always be followed

How to Use This Knowledge Base

Once the template above is populated, here is how it plugs into your workflow.

For Claude (Primary)

The completed knowledge base will be loaded as a Claude Project. This means every conversation inside that project automatically has the knowledge base as context. You open the project, ask Claude to write a cold email or a LinkedIn post, and it already knows your voice, your ICP, your regulations, and your rules.

Nandesh will set up separate Claude Projects for different use cases:

Arcus Content Studio -- for LinkedIn posts, articles, thought leadership

Arcus Outreach -- for cold emails, LinkedIn DMs, follow-up sequences

Arcus Client Comms -- for proposals, client emails, reports

Arcus Compliance KB -- for regulatory research, client advisory, compliance content

Libra -- (if separate brand, gets its own project)

Each project will reference the relevant sections of the knowledge base.

For ChatGPT (If Still Used)

If you keep ChatGPT for specific tasks, you can paste relevant KB sections into a Custom GPT's instructions. But the goal is for Claude to be primary.

For Any Future Tool

Because the knowledge base is structured and lives in Notion, any AI tool that can read from Notion (or receive pasted context) can use it. The knowledge base is tool-agnostic by design.

Validation Checklist

After the knowledge base is generated, Lee reviews it against this checklist.

[Unsupported block type: to_do]

What Happens Next

Lee completes the ChatGPT extraction (Migration Worksheet)

Outputs are sent to Nandesh

Nandesh runs the outputs through Claude to populate this template

Lee reviews the populated knowledge base for accuracy

Nandesh creates Claude Projects with the knowledge base loaded

Lee tests by asking Claude to write a cold email, a LinkedIn post, and a client email

Lee flags anything that feels off

Nandesh refines the knowledge base based on feedback

Knowledge base is locked as v1.0

All future content and messaging runs through Claude with this context

The knowledge base is a living document. As Arcus evolves, regulations change, or Lee's positioning shifts, the KB gets updated. But v1.0 should cover at least 90% of what is needed for day-to-day content and messaging.

Pulled live from Notion via official API · rendered by custom React · cached 5 min · updates on next visit after page edit.