The AI Daily Brief: Artificial Intelligence News and Analysis

Episode: How to Build a Personal Context Portfolio and MCP Server
Host: Nathaniel Whittemore (NLW)
Date: April 3, 2026

Episode Overview

In this episode, NLW explores the concept of building a "Personal Context Portfolio" (PCP)—a portable, machine-readable guide to your identity, preferences, work, and expertise designed for integration with AI agents and tools. He explains why context is crucial for effective AI interactions, how organizations and individuals struggle with context management, and offers a hands-on framework for creating your own PCP. The episode caps off with practical steps for setting up your portfolio and deploying it on a Minimal Context Protocol (MCP) server, ensuring smooth, context-rich AI experiences without repeating yourself endlessly.

Key Discussion Points & Insights

1. The Importance of Context in an Agentic Era

• Agents Need Deep Context:
  In the age of agentic AI (autonomous, task-oriented digital agents), providing rich, accessible context is essential for effective collaboration and results.

  “Right now we officially live in the agentic era and agents as we know, need context to do their jobs well. And yet context is one of those things that is very simple to articulate and much harder to actually organize in a way that is useful.” (03:20)

• Enterprise Struggles:
  Organizational data is rarely AI-ready; most companies have messy, unstructured info, posing a barrier for agents to be truly useful in business.

  • NLW cites Michael Chen (Applied Compute):

    "Data ready is just a state of mind. The gap between we have data and we have data in a format that an AI system can learn from is enormous." (05:15)

• Context ≠ Data:
  While often used interchangeably, context is broader than just data—it’s what allows AI to tailor its outputs and actions to user needs.

2. Current Solutions & Gaps (Enterprise and Individual)

• Enterprise Solutions:
  Organizations like Notion are building “librarian” agents managing organizational context, and Andrew Ng introduces context-sharing tools for coding agents.

  “Notion … basically their entire play for enterprise AI is a pitch that they already have your enterprise's context … a team of little librarians in your database.” (06:23)

  “Andrew Ng recently wrote, should there be a stack overflow for AI coding agents to share learnings with each other?” (07:21)

• Individual Context is Left Out:
  There’s little focus on making personal context portable between AI tools—your agent “remembers” you on one platform, but not when you switch to another.

• Recent Industry Developments:

  • Anthropic (Claude) introduced a basic way to import “memories” from ChatGPT, but it’s simplistic: a prompt that tries to export what one agent knows about you for another agent.

    “Claude's approach to importing memory was pretty simplistic… basically, it was a prompt that asked ChatGPT to write up everything it knew about you so you could hand that document off to a new Chatbot.” (10:00)

3. The Personal Context Portfolio (PCP): Vision and Structure

• Problem & Solution:
  Constantly re-explaining yourself to new agents is time-consuming and loss-prone (“context repetition tax”). The solution: a structured, modular set of markdown files that any AI tool can read.

• Design Principles:

  1. Markdown-First: Universal, machine-readable.
  2. Modular: Separate files for different dimensions of context.
  3. Living, Not Static: Designed to be updated as you grow or change roles.
  4. Portability: Easily used across any platform or agent.

• Portfolio Template: The 10 Core Files

  1. identity.md: Name, role, organization, a distilled “about you.”
  2. rolesAndResponsibilities.md: What you actually do day to day.
  3. currentProjects.md: Active projects, goals, collaborators, status.
  4. teamAndRelationships.md: Key people, roles, interaction dynamics.
  5. toolsAndSystems.md: What apps you use, your setup, connections.
  6. communicationStyle.md: Preferences for language, tone, format.
  7. goalsAndPriorities.md: Big-picture goals, what you’re optimizing for.
  8. preferencesAndConstraints.md: “Always do/never do” rules, restrictions.
  9. domainKnowledge.md: Areas of expertise, specialized terminology/context.
  10. decisionLog.md: Past decisions with rationale/history.

  “Effectively, it's an operating manual for any AI that works with you… It’s API documentation, but for you, a single source of machine-readable truth about who you are.” (15:50)

4. How to Build Your PCP Efficiently

• AI Interview Loop:
  Don’t do it by hand! Use an AI (Claude, ChatGPT, Gemini, etc.) to interview you and create draft files.

  • Steps:
    1. Initiate a project in your chatbot of choice.
    2. Go through iterative interviews (draft → react → revise) for each file.
    3. Use the provided templates and protocols, available through the podcast’s GitHub repo and companion app.
  • NLW’s PCP app:
    • Two sections: the ongoing interview (powered by Opus 4.6) and the portfolio being built.
    • Answers can update multiple files at once.
    • Download your portfolio anytime; it’s private and free.

  “Rather than having to break this up into 10 different interviews… when you answer one question, if it's relevant for different portfolio files, that's all going to be added at once.” (38:11)

• Resources:

  • Public GitHub repo: Templates, interview protocols, and synthetic demos (entrepreneur, executive, knowledge worker).
  • “Wiring” folder: Resources for turning this portfolio into a Claude project, MCP server, or API layer.

5. Taking it Further: Deploying to an MCP Server

• What’s an MCP Server?

  • A program that exposes your context files via a standard protocol. Agents can ask, “what do you have?” and request files as needed.

    “A program that responds to a specific protocol. An AI tool sends it a request saying, what do you have? And it responds with a list of resources. The tool says, give me this resource, and it responds with the content.” (50:10)

• Process:

  1. Decide if you want local or remote access (sometimes both!).
  2. Use AI as your “build partner”—walk through setup step-by-step.

     “AI is zero judgment. There is no risk of you looking or seeming dumb, because there's no one on the other end…” (52:43)

  3. Troubleshoot together when errors arise (file naming, port conflicts, etc.).
  4. To go remote:
     • Create a GitHub repo.
     • Copy your portfolio files.
     • Adjust server code.
     • Deploy (e.g., via Railway).
  5. Most of the work is troubleshooting, but overall, setup is lightweight.

  “The jump from local MCP server to something that was available on the web actually took less time than local just because we ran into fewer issues.” (01:03:00)

6. Learning by Doing

• Hands-on Benefit:
  Even if you don’t plan to use MCP servers long-term, you’ll gain value just by defining your own context files and practicing structured handoff of information to agents.

  “A lot of the value you’re going to get out of the follow along for this is going to be just in the creation of the files.” (01:05:20)

Notable Quotes & Memorable Moments

• On the need for portable, structured personal context:

  “As you get into the world that we're going into, where every week there are going to be new types of agents and agentic surfaces that you're interacting with, it is going to become absolutely critical to have a way to get out of paying this context repetition tax.” (13:15)

• On modular, living portfolios:

  “This is not a thing you write once, but it’s a thing you maintain or better, that your agents help you maintain. As projects change and priorities shift, the personal context portfolio should evolve with you.” (19:50)

• On AI as a nonjudgmental tutor:

  “AI is zero judgment. There is no risk of you looking or seeming dumb, because there's no one on the other end of the line to think that when you are trying to get things explained step by step…” (52:43)

Timestamps of Key Segments

• 03:20 – Need for context in the age of agentic AI
• 05:15 – Data readiness challenges in enterprises (Michael Chen/Applied Compute)
• 06:23–07:21 – Notion’s “context librarians” and Andrew Ng’s “Context Hub”
• 10:00 – Claude's "export your memory" feature and its limitations
• 15:50 – Introducing the Personal Context Portfolio concept
• 19:50 – Design principles of the PCP
• 24:45 – Walkthrough of the 10 core PCP files
• 38:11 – PCP Agent/App and advantages of simultaneous file updates
• 50:10 – What is an MCP server?
• 52:43 – Using AI as a hands-on tutor for setup
• 01:03:00 – Local vs. remote MCP server deployment lessons
• 01:05:20 – The intrinsic learning value of building your PCP

Actionable Resources

• GitHub Repo: Templates, protocols, and example files for your Personal Context Portfolio
• PCP Interview Agent App: Automated, privacy-first tool to auto-generate your portfolio
• Companion Experiences: Available via Play.aidailybrief.ai (see episode notes for exact links)

Summary

This episode equips listeners with both the “why” and “how” for building a truly portable, machine-readable “vCard for your digital life.” NLW demystifies the rising problem of “context tax,” offers a tangible template (in markdown) for personal context, and guides you through both the creation and operationalization of your portfolio—including deploying it on an MCP server for agentic interoperability. By tapping the latest in agent design and AI tooling, listeners are invited to transform how they onboard, manage, and collaborate with AI, ensuring future agents always have the full context they need—no more repetition, no more lost nuance, just seamless, personalized interaction.

Listen to the full episode for walkthroughs, commentary, and direct links to tools and resources.