source system overview

what this project is

this project is building a private ai chief of staff for a high-agency principal.

the product is not a chatbot and not a "second brain."
it is a private operating layer that holds context, priorities, relationships, and approvals close to the principal and turns them into prepared action.

why this source system exists

we do not want future agents, operators, or collaborators to guess where the doctrine came from.

this source system makes three things explicit:

1. which sources shaped the product
2. what we actually took from each source
3. which parts affect pitch, system design, and future agent behavior

source families

1. egor rudi conversation

this is the strongest first-party source on what the product should feel like.

it gives us:

• the shift from second brain to ai chief of staff
• the first wow: secure all-access plus immediate leverage
• the bar for situational awareness
• the demand for quick wins over intelligence theater
• the idea that the system should update the user's mental model of its own capabilities

public detail page:

• https://ai-chief-of-staff-egor-rudi-source.vercel.app

2. mitchell levin transfer

this is the strongest source on runtime logic and control coherence.

it gives us:

• objectives as setpoints
• memory as active control state
• nested agency
• horizon-aware agents
• topology-triggered memory reconciliation
• perturbation-first evaluation

public detail page:

• https://ai-chief-of-staff-mitchell-levin-so.vercel.app

3. daniel miessler source family

this is the strongest source on category language and infrastructure framing.

it gives us:

• scaffolding over model
• personal ai infrastructure
• inspectable memory
• desired outcome management
• private operating layer as the right category

public detail page:

• https://ai-chief-of-staff-daniel-miessler-s.vercel.app

how the three sources combine

egor gives the product feel and the buyer's truth.

levin gives the control logic and runtime discipline.

miessler gives the category language and infrastructure doctrine.

put simply:

• egor tells us what must feel true
• levin tells us how the system must stay coherent
• miessler tells us how to frame and scaffold it

what future ai agents should understand

any future agent working on this repo should assume:

• positioning comes from first-party experience, not abstract category games
• system design must preserve trust, auditability, and bounded action
• memory is not storage; it changes future action selection
• chat is only one surface; task-native interfaces matter
• the system wins by prepared clarity, not by sounding smart

current weak points

• no dedicated memo yet for the ro / r0 source family if that source is real and distinct
• the first wow flow is conceptually clear but not yet frozen as a step-by-step spec
• public pages exist as readable summaries, but the internal local source pack remains richer than the public layer