For Chief of Staff networks & ops leaders

Your company's living knowledge base, maintained by an agent.

Automatically turn your Slack, email, and docs into an LLM wiki with granular permissions — queryable from Slack, iMessage, or your own Claude.

The problem

Your institutional memory lives in Slack, and nobody's reading it.

Every decision, customer insight, and half-built process is buried in a firehose of messages. Company wikis go stale in weeks. Onboarding a new hire is a scavenger hunt through DMs. And the moment anyone leaves, their context walks out the door.

✕ Manual wikis

Stale within weeks. Nobody writes them. Nobody updates them. The one person who knew the system left last year.

✕ Slack search

Returns 400 messages, none of which answer the question. Context is scattered across channels, DMs, and threads.

✕ Notion / Confluence

Good primitives, but agents can't write to them at ingestion time with appropriate visibility. Still manual.

✓ Agent-maintained LLM wiki

Slack history becomes a structured, searchable knowledge base with per-item permissions the agent sets as it ingests.

Data sources

We ingest the systems where decisions actually happen.

Slack first — it's where most modern orgs think out loud. Email, docs, and personal notes fill in the gaps.

Slack Primary

Full-history ingest plus a quiet bot in-workspace. Extracts decisions, ideas, relationships, customer context. A "brain" reaction flags anything worth saving — or the agent figures it out on its own.

Email Optional

Captures external conversations — customers, partners, investors — that never reach Slack. Feeds the same unified graph the agent queries.

Google Drive Optional

Docs, decks, spreadsheets connected to Slack/email context. Answer "where's the latest SOP?" or "what did the board deck say about this customer?" in one query.

Desktop / local files Power users

For Chiefs of Staff and founders who keep important notes locally. Indexed into the same wiki your team sees, with private-by-default visibility.

How it works

Raw activity → structured LLM wiki → agents on every surface.

Four stages. The middle two are where the magic happens — and where per-item visibility is the unlock nobody else has.

1

Ingest activity

Import Slack history and watch incoming messages. Optionally connect email, Drive, and local files. Nothing moves to a new format — your source systems stay the source of truth.

2

Build the LLM wiki / knowledge graph

An agent extracts entities (people, companies, projects, customers) and relationships between them. Builds profiles, SOPs from recurring patterns, and a corporate idea bank. Everything lives as structured markdown in a git-backed repo you own.

3

Apply granular permissions — at ingestion time

This is the unlock. The agent classifies each extracted item and sets its visibility as it writes: team-private, department-only, company-wide, or public. No manual tagging. No post-hoc cleanup. No sensitive thread leaking to the whole org.

4

Deploy agents on every surface you use

Slack bot in your workspace. iMessage / WhatsApp as a control surface for on-the-go queries. Headless Claude or GPT chat with the wiki as context. All three hit the same knowledge base with the same permissions.

The insight that makes this possible

Every "company wiki" tool fails the same way: you can't separate public from private at the speed information gets created. One sensitive thread poisons the whole wiki. Manual tagging doesn't scale. Notion's per-page sharing is clunky for an agent writing thousands of extractions a day.

ListHub is the first system where the agent sets visibility at write time as a first-class field on the API. The same primitive that lets a personal LLM wiki selectively publish its research notes while keeping its diary entries private — applied at company scale.

Agent surfaces

Three surfaces. One knowledge base. Same permissions.

Different surfaces for different moments. All of them hit the same underlying wiki and respect the same visibility rules per user.

Slack bot

Primary collaboration surface

Lives in your workspace. Quiet by default — doesn't spam every channel. Auto-saves decisions and ideas.

  • Mediates discussions with context
  • Surfaces relevant history when asked
  • Marks messages with a "brain" reaction
  • Maintains an idea bank automatically

iMessage / WhatsApp

Control surface

Text the agent for quick answers. This is a remote control, not a group chat participant.

  • "What did we decide about X?"
  • "Summarize this customer's history"
  • "Who owns this initiative?"

Headless Claude / GPT

Power user path

The agent uses your company wiki as context automatically. Works with any AI chat client you already use.

  • Deep analysis & drafting
  • SOP generation
  • Automation opportunity scans
Use cases

Where this lands first.

Four concrete shapes this takes in the wild. The first one is where we're starting.

Chief of Staff networks & AI-first orgs

Automatic company knowledge base creation from Slack and adjacent systems. Onboards new CoS and execs 10× faster. Keeps institutional memory current without anyone maintaining a wiki.

Process capture, SOPs, and automation

Like Scribe but passive. Capture processes as they happen in Slack and calls. Turn recurring patterns into SOPs or automation candidates. No more stale "how we do X" docs.

Idea banks and innovation management

The bot maintains an idea bank automatically for a 100–1000-person org. A single-function value prop for innovation officers: "we maintain your idea bank automatically."

Personal + organizational wikis, unified

Your personal LLM wiki (private by default, selectively public) plus the company wiki, in one tool with one permission model. The same primitive scales from one person to a company.

Who this is for

Early design partners.

We're working directly with a small number of organizations to shape the product. If you're in either group below, we want to talk.

Primary fit

Series A–B startups, 30–100 people

Big enough to have knowledge sprawl and alignment problems. Small enough to adopt new workflows without six weeks of change management. Usually led by an ops person or CoS who's been quietly wanting this for a year.

Secondary fit

Chief of Staff networks & transformation leads

If you run or participate in a CoS network exploring how much of the role can be augmented or automated, this is the substrate you've been missing. Built for AI-first operating models from day one.

Interested?

We're onboarding a handful of design partners. If your team wants their company knowledge working for them instead of against them, get in touch.

✉️  [email protected]

Or poke around the underlying platform at listhub.globalbr.ai — everything on this page is built on its per-item visibility primitive.