mozart // entry offer  ·  architecture one-pager  ·  draft v0.1

The Company Second Brain a hosted, agent-accessible knowledge layer companies install once and build everything on top of

status: side-quest / pre-build  ·  2026-07-12  ·  owner: TG  ·  not indexed

Take Andrej Karpathy's LLM Wiki method (a knowledge base an AI builds and maintains for you, not one you maintain and occasionally ask about) and turn the single-player laptop version into a hosted, multi-tenant, MCP-accessible company brain on Cloudflare. Mozart installs it, caters the structure to the business, wires it to their tools, and maintains it over time.

the wedge // every writeup of Karpathy's method is "build your own personal one in Obsidian in an hour." Nobody has made the team version: hosted, private, secured, and reachable by both humans and their AI agents through one MCP endpoint. That gap is the product. We are not competing with Notion, we are building the data layer that sits underneath a company's agents.

why this is the right entry offer for Mozart

foundational

Every workflow Mozart builds afterward needs a source of truth to read from and write to. Install the brain first and every later build gets faster and cheaper.

land & expand

Low-risk, high-trust first project. "Let us build your company brain" is easier to say yes to than a big automation. It opens the door to the workflow revenue behind it.

defensible

Once a company's knowledge lives in a brain Mozart installed and maintains, Mozart is wired into the operating core. That is a moat, not a one-off deliverable.

the three layers (Karpathy's model, hosted)

1 · raw append-only, immutable, the AI reads but never edits

Everything the company ever feeds in, untouched: call transcripts, docs, PDFs, Slack threads, emails, web clippings, meeting notes, screenshots, raw thoughts. The single source of truth of what has been seen. Stored in R2, catalogued in D1.

2 · wiki AI-owned, structured, self-maintaining

Encyclopedia-style markdown the AI writes and keeps consistent: entity pages (people, accounts, products), concept pages, summaries, comparisons, an INDEX.md, and backlinks between everything. The AI creates pages, updates them when new raw arrives, fixes cross-references, and resolves contradictions. Rendered from R2/D1, embeddings in Vectorize.

3 · outputs auditable answers & reports

Synthesized responses to queries and scheduled digests, saved as durable records (YYYY-MM-DD-query-slug.md). This is where "what did the brain tell us and when" lives, so answers are traceable, not ephemeral.

Governed by a per-company schema file (Karpathy's CLAUDE.md idea): ingestion rules per source type, index format, backlink + naming conventions, lint rules, and how to resolve conflicting sources. This is the file we cater to each business and the core of the "install."

company-brain/ ├─ SCHEMA.md # per-company rules the AI obeys (the catered install) ├─ raw/ # immutable sources → R2 + D1 catalog │ ├─ calls/ docs/ email/ slack/ web/ notes/ ├─ wiki/ # AI-owned knowledge → markdown + Vectorize embeddings │ ├─ INDEX.md │ ├─ entities/ # people · accounts · products · vendors │ ├─ concepts/ # how-we-do-X · policies · playbooks │ └─ summaries/ ├─ outputs/ # answers + digests → YYYY-MM-DD-slug.md └─ inbox/ # intake queue awaiting sync

the daily loop (four operations)

01

capture

A source hits inbox/: via MCP, an email address, a webhook, or a tool sync. Nothing is organized yet, it is just safely captured.

02

sync

The AI reads INDEX.md, extracts entities/concepts, creates or updates wiki pages, writes backlinks, updates the index. Runs on a Cloudflare cron + on-demand.

03

lint

Health pass: find contradictions, stale pages, duplicates, orphaned links, and concepts referenced but missing. Flag or auto-fix. This is what keeps trust high.

04

digest

Scheduled rollup of what changed and what is worth a human's eyes. Pushed to Slack/email. Keeps humans in the loop without them doing the filing.

Karpathy's line, and the whole reason this works: humans abandon wikis because upkeep grows faster than value. LLMs do not get bored, do not forget a cross-reference, and touch 15 files in one pass.

the cloudflare stack

R2raw source files + rendered wiki markdown. Cheap, private, no egress fees.
D1the spine: source catalog, the page/link graph, the intake queue, per-tenant metadata, audit log. (Structured retrieval only, not semantic.)
Vectorizeembeddings for semantic search. This is the piece a wiki-on-D1-alone is missing, and it is what makes "ask the brain anything" actually work.
Workers + Workflowsthe sync / lint / digest jobs, ingestion connectors, and every automation Mozart builds afterward. Durable, retryable, scheduled by cron.
Workers AI / modelembeddings + the read-write reasoning that maintains the wiki. Model-agnostic.
MCP server (Worker)the one front door. Humans and agents (ChatGPT, Claude, a Hundo-style agent) query and write the brain through it. Per-tenant auth + scoping.
Access / Turnstilesecure the tenant boundary. Each company's brain is isolated and private, which is the whole reason they will not use Notion for this.

intake: how data actually gets in

push

Dedicated inbox email, webhook URL, and an MCP capture tool so a human or agent can drop anything in from anywhere.

pull / sync

Connectors that sweep the company's tools on a schedule: Gmail, Drive, Slack, calendar, CRM, call recorders. Each becomes raw sources.

the MCP surface (humans + agents, same door)

A small, safe tool set exposed to the company's team and their agents:

search(query) ask(question) → cited answer capture(source) get_page(id) list_recent(entity) digest(range)

Because it is MCP, the same brain answers a person in Claude, a teammate in ChatGPT, and an autonomous agent mid-workflow. That universality is the "why now."

go-to-market (inside Mozart)

entry offer

"Install your company second brain." Fixed-scope, fast, catered schema + tool connections.

retainer

Maintain, monitor, tune the loop, add sources over time. Recurring revenue, ongoing trust.

expand

Every workflow after this reads/writes the brain. Faster to build, easier to sell, higher ceiling.

the honest risks (where these die)

open questions for TG

1. First install: a warm Mozart customer, or your own companies (100.partners / Mozart) as the dogfood case first?
2. Day-one scope: flat whole-company brain (fast), or per-role permissions from the start (slower, more enterprise-ready)?
3. Pricing shape: one-time install + monthly retainer, or bundle the brain into a larger Mozart engagement as the "phase 0"?