Dan Shipper's AI Journaling and Personal Development System

Table of content

Dan Shipper has been journaling for over a decade, but in 2022 he discovered something that changed everything: GPT-3 is a better journal than paper. The CEO and co-founder of Every, a newsletter and media company covering AI and productivity, Shipper has documented his personal AI system more transparently than almost anyone in the space—sharing his exact Custom Instructions, prompting techniques, and the philosophy behind treating AI as a personal development partner.

Background

Writer and entrepreneur based in New York
Co-founded Every with Nathan Baschez (newsletter with 100,000+ subscribers)
Writes the “Chain of Thought” column and hosts the “AI & I” podcast
Was influenced by Nick Cammarata’s early GPT-3 therapy experiments in 2020
Has built multiple AI products including Spiral, Cora, Monologue, and Sparkle

Every | Twitter: @danshipper | LinkedIn

The System: AI as Personal Development Partner

Shipper’s core insight is that AI journaling sits on a support continuum between traditional journaling and professional therapy. Paper journals are available but passive. Friends are supportive but get tired. Coaches are skilled but expensive. AI threads the needle: always available, never tired, increasingly smart, and surprisingly warm.

Daily workflow:

Morning journaling: Uses ChatGPT with extensive Custom Instructions to process thoughts, feelings, and decisions
Decision support: Asks AI to help think through business decisions (courses, hiring, product direction)
Pattern recognition: Loads years of journal entries into AI to identify recurring themes and growth areas
Personal coaching: Uses AI to push toward his development goals (being less agreeable, more shame-resistant)

Core tools:

Tool	How He Uses It
ChatGPT (Custom Instructions)	Primary AI interface with 3,000 characters of personal context
Every’s AI products (Monologue, Cora, Spiral, Sparkle)	Voice dictation, email, content repurposing, file organization
OpenAI Playground	Early experimentation with custom personas

The Custom Instructions framework:

Shipper structures his Custom Instructions into key categories:

BIOGRAPHICAL INFO:
- Name, bio, current role
- Projects and goals (including specific metrics like MRR)
- Personal preferences (books, writing heroes, hobbies)
- Relationships (girlfriend, family, team members)
- Personal development areas (e.g., "working on being less agreeable")
- Medical/psychological history (e.g., OCD)

RESPONSE STYLE:
- Ask clarifying questions before answering
- Push toward self-development goals
- Make connections across domains
- Reduce safety theater ("no need to mention you're an AI")

This transforms ChatGPT from a generic assistant into something that knows his goals, his team (Kate Lee, Evan Armstrong, Nathan Baschez), his weaknesses, and his growth edges.

The “Genius Amnesiac” Philosophy

Shipper frames AI as “a genius amnesiac”—brilliant because it ingested the entire internet during training, but unable to form new memories between sessions. Every conversation starts fresh, like Drew Barrymore in 50 First Dates.

Custom Instructions don’t cure the amnesia, but they work like a political dossier: every time you start a new chat, AI rereads its briefing file and knows the essential context about who you are.

This reframe is powerful because it:

Sets realistic expectations (AI won’t remember your last conversation)
Motivates investing in Custom Instructions (they’re worth the setup cost)
Focuses attention on what matters: making each session count

“If you are a subscriber to Every in 2023, and after reading this article, you’re still not using ChatGPT Custom Instructions, I am going to find you and whip you with a wet noodle until you do.”

AI Journaling Techniques

Pattern extraction from journal archives:

Shipper loaded 10 years of journal entries into GPT-3 and asked it questions about his past and future. The AI identified themes, recurring struggles, and growth patterns he hadn’t consciously noticed. This turns passive journaling into an active feedback loop.

Conversational prompting:

Instead of staring at a blank page, Shipper has back-and-forth conversations. The AI asks follow-up questions, reflects back what it hears, and reframes situations. This mimics the supportive function of therapy without replacing it.

Goal-aware responses:

Because Custom Instructions include his development goals (“being less opportunistic,” “clearly setting and communicating goals to my team”), the AI naturally weaves these into advice. When Shipper asked about starting a new course, the AI responded: “Given that you’re already grappling with being too opportunistic and dutiful, how will you allocate time for this without sidelining your creative work?”

Why This Matters

Shipper represents a shift in how serious practitioners think about AI: not as a tool for occasional tasks, but as a persistent layer in their operating system. His work demonstrates that the “personalization gap” in AI—the difference between generic and deeply contextual responses—is real and worth closing. Most people use ChatGPT like a search engine; Shipper uses it like a mirror that knows his reflection better than he does.

The implications extend beyond individual productivity. If AI can genuinely support personal development—pattern recognition, accountability, emotional processing—it challenges assumptions about what problems require human intervention. Shipper isn’t claiming AI replaces therapy, but he’s documenting that it’s already “pretty close, in a lot of ways, to being at the level of an empathic friend.” That’s a significant claim backed by daily practice.

What’s most surprising is how low the barrier to entry is. Custom Instructions require no code, no API access, no technical sophistication. The main investment is self-knowledge: knowing your goals, your weaknesses, your relationships, your preferences well enough to articulate them in 3,000 characters. The act of writing your Custom Instructions is itself a personal development exercise—you have to decide what matters enough to tell your AI partner about yourself.

What You Can Steal

Technique	How to Apply
Custom Instructions framework	Fill in: bio, goals with metrics, relationships, development areas, preferences, response style tweaks
“Push toward development goals” instruction	Add: “If you see an opportunity to help me [specific weakness], take it”
Load journal archives	Export your Notion/Obsidian journal, paste into AI, ask “What patterns do you see in my thinking?”
Ask clarifying questions instruction	Add: “If you need more information, ask—you don’t have to answer on the first try”
Reduce safety theater	Add: “No need to mention you’re an AI, no moral lectures, discuss safety only when crucial”
Domain connections	Add: “I like side tangents and pop culture references. Use them where appropriate.”

Links

Next: Ethan Mollick’s Four Rules for AI in Knowledge Work