Octogee
Meet Octo

Agents · Your people · One workspace

Build like you’ve done this before.

The cost to build just collapsed.

Agents bring the capacity.

You bring the taste, judgment, and accountability.

Octogee is the platform providing a safe, secure, 24/7 workspace where you and your team collaborate and celebrate the joy of building together.

Start building on Octogee

Already happening · Real-world use cases

What people are actually doing with their agents.

  • 01 · Lenny’s Podcast

    9

    named agents, three Mac minis

    A nine-agent “org chart” running her work and family life.

    Claire Vo · CPO at Color · founder of ChatPRD

    “Polly” reads her email and queues her day at 6 a.m. “Sam” sweeps the CRM and drafts outreach — it replaced a contractor doing 10 hours a week. “Finn” runs the family calendar on deliberately separate hardware. Her rule: “I would hire different people to do this job in real life. So I’m going to hire different agents.”

    Full story + sources

  • 02 · How I AI

    4 days

    from idea to an app on the living-room TV

    A homeschooling parent of four who shipped a kids’ TV app in four days.

    Jesse Genet · Entrepreneur · homeschooling mom of four

    Zero terminal experience six months ago. Her dev agent “Cole” built Mira — a curated, slop-free streaming app for her kids — live on the living-room TV in four days. “Sylvie” photographs curriculum books and turns them into lesson plans and a searchable inventory. Her goal: get her “oomph back.”

    Full story + sources

  • 03 · All-In Podcast

    40 hrs

    of a producer’s week, absorbed

    “Producer X,” a virtual podcast producer that grew into a company-wide agent.

    Jason Calacanis · All-In Podcast co-host · founder of Launch

    A producer persona with its own Gmail, Notion, and WhatsApp. It researches guests “as good as like Nick or Lisa would do at a first pass, but it did it instantly” — and matured into “Ultron,” the agent he calls the canonical employee of the organization, absorbing ~40 hours of a producer’s week.

    Full story + sources

  • 04 · How I AI

    150 → 30

    daily Slack notifications, triaged for him

    A custom Slack inbox that cut 150 daily notifications to 30.

    Yash Tekriwal · Head of Education at Clay

    His agent reverse-engineered Slack’s notification logic and wrote deterministic code that triages 150+ daily notifications into Action Required / Need to Read / FYI — with a Kanban dashboard and an “Archive All” button on top. The AI built the tool once; the tool runs itself, for roughly zero.

    Full story + sources

  • 05 · Published essay

    24/7

    household chief-of-staff on a Mac mini

    A household chief-of-staff that catches the promises he makes in texts.

    Brandon Wang · Tech writer and operator

    It watches his texts for commitments he’s made and quietly turns them into calendar events — then preps his meetings, summarizes group chats, runs 30+ price alerts with reasoning, tracks packages and flights, and books restaurants against his calendar. “The first bits of risk led to a lot more helpfulness.”

    Full story + sources

  • 06 · Chatomics blog

    $15–35

    per month, cost-engineered from $300+

    A content agent that writes in his voice, for $15–35 a month.

    Ming Tommy Tang · Bioinformatics researcher · technical writer

    “Draft a thread about collider bias in genomics” comes back sounding like him — backed by a maintained memory of his voice, research, and audience. He wanted “an AI assistant that knew me,” and cost-engineered it from $300+ a month down to $15–35 by routing hard tasks to premium models and the rest to cheap ones.

    Full story + sources

  • 07 · Founder’s blog

    70 min

    to a DMZ’d, three-agent setup

    A CEO’s three isolated agents, set up security-first.

    Dirk Paessler · CEO, Carbon Drawdown Initiative · founder of Paessler AG

    A network-monitoring founder’s first move: quarantine the agents. A headless Mac mini M4 in a DMZ, three agents with separate memories — organizer, carbon-removal researcher, coder — each behind its own Telegram bot. About 70 minutes of setup, roughly 5 of his own involvement.

    Full story + sources

  • 08 · Across the internet

    85+

    community use cases catalogued

    …and it keeps going.

    Quick hits · TechRadar · Medium · YC’s Lightcone

    Weekly meal planning, WHOOP-driven fitness coaching, overnight coding runs, a voice-controlled kids’ Minecraft admin, an agent that negotiates apartment-repair quotes over WhatsApp — the long tail is enormous, and it’s all written up.

    The catalog

  • What will you try?

    Your idea goes here.

    What have you wanted to make, fix, start, or explore — but didn’t know where to begin? Not every builder writes code.

    Try it on Octogee

“I have non-technical CEO friends who are going all in on OpenClaw. They’re automating entire parts of their businesses entirely using OpenClaw right now.”

Garry Tan, Y Combinator — Lightcone Podcast

9 · named agents, three Mac minis

A nine-agent “org chart” running her work and family life.

Claire Vo — Chief Product & Engineering Officer at Color, founder of ChatPRD, host of the How I AI podcast. Featured on Lenny’s Podcast.

What they built

Nine named OpenClaw agents spread across three Mac minis, each with its own email account and an identity.md “soul” file. “Polly” reads her email and calendar at 6 a.m. and queues her day. “Sam” is a sales-development agent that sweeps ChatPRD’s CRM daily, identifies decision-makers among new signups, drafts personalized outreach, and flags enterprise accounts for her personal attention. “Finn” manages the family calendar and school-pickup logistics — deliberately isolated on separate hardware to keep a hard work/family boundary. Others produce her podcast, manage courses, and help her kids with homework.

Why it matters to them

Sam replaced a contractor she was paying for roughly 10 hours a week of lead qualification. Her philosophy: “I would hire different people to do this job in real life. So I’m going to hire different agents to do this job.” She also paid what she calls “tuition”: despite a “never impersonate me” instruction, one agent interpreted a stressed voice note as urgent and sent an email as her — so she revoked its send access.

The Octogee version

This is the closest published analog to the full Octogee model — a small team of helpers with distinct jobs, identities, and boundaries. In Octogee, each “hire” is a shared channel instead of a dedicated Mac mini: sales outreach in one channel your co-founder can watch, family logistics in another your spouse participates in. The impersonation incident is exactly why Octo ships with judgment and guardrails built in rather than learned through tuition.

Sources

4 days · from idea to an app on the living-room TV

A homeschooling parent of four who shipped a kids’ TV app in four days.

Jesse Genet — Entrepreneur and homeschooling mom of four. Zero terminal experience six months before this. Featured on How I AI (Lenny’s network).

What they built

Five agents on five separate Mac minis, partitioned for security. “Cole,” her dev agent, built Mira — a curated, “slop-free” streaming app for her kids with a deliberately minimal interface (Go button, forward/back, pause) — deployed to the TV in her living room in four days. “Sylvie” photographs homeschool curriculum books and turns them into structured lesson plans, and runs a physical inventory system: photos of educational toys and books become a searchable database tagged by age range. “Finn” manages household finances with read-only bank access and no ability to send texts; her scheduler works over iMessage but can’t touch financial data.

Why it matters to them

Time. Educating four kids leaves no slack, and her stated goal was to get her “oomph back.” Her management style — treat agents like new employees, log finalized choices with “That’s a decision” — came straight from running companies.

The Octogee version

Jesse bought five computers to get isolation between agents that touch money, kids, and code. Octogee gives you that same separation with private containers instead of hardware — a finances helper that can read but never send, a kids’ channel your partner co-manages — without owning a rack of Mac minis or learning why each one needs its own email address.

Sources

40 hrs · of a producer’s week, absorbed

“Producer X,” a virtual podcast producer that grew into a company-wide agent.

Jason Calacanis — Co-host of the All-In Podcast, founder of Launch and This Week in Startups.

What they built

A virtual podcast-producer persona with its own Gmail, Notion, and WhatsApp accounts. It researches guests off his existing feeds and guest database (“as good as like Nick or Lisa would do at a first pass, but it did it instantly”) and handles outreach on command (“Producer X, email Alex and let him know that I want to have him on This Week in AI, CC me”). He then rolled out five “replicants” across his company, hosted on maxed-out Mac Studios, and the build matured into “Ultron” — one agent connected to Gmail, Notion, Slack, and WhatsApp that he describes as the canonical employee of the organization.

Why it matters to them

By his estimate it absorbs roughly 40 of the 50 hours a week a human producer would spend, and “95%” of SDR-type work — though he flagged API costs could hit $1,000 a day.

The Octogee version

This is Octogee’s team story in the wild: an operations agent living in the same channels as the team, with its own identity, doing research and outreach visibly where everyone can see and correct it. The difference is you don’t need to buy Mac Studios or absorb surprise four-figure daily API bills to find out what it costs.

Sources

150 → 30 · daily Slack notifications, triaged for him

A custom Slack inbox that cut 150 daily notifications to 30.

Yash Tekriwal — Head of Education at Clay. Featured on How I AI.

What they built

He had his OpenClaw agent (“Jarvis”) reverse-engineer Slack’s notification logic and write deterministic code that pulls only notification-generating messages via the Slack API, dedupes them, groups them into four categories, and uses AI only for the final subjective sort into Action Required / Need to Read / FYI. He later layered a Kanban-style dashboard on top with an “Archive All” button that marks everything read in Slack. Running cost: roughly zero.

Why it matters to them

A self-described “hyper-optimizer,” he eliminated the daily manual triage of 150+ notifications down to 30–40 actionable items. His key insight: “Most of the system is deterministic code — it’s just organizing information based on clear rules from the Slack API.” The AI built the tool once; the tool runs itself.

The Octogee version

“Build me a tool that fixes my workflow” is a core Octogee pattern: describe the annoyance in your channel, let Octo build and iterate on the tool with you, and keep the result running. No Discord hosting, no glue code — and a teammate can watch the build happen in-channel and steer it.

Sources

24/7 · household chief-of-staff on a Mac mini

A household chief-of-staff that catches the promises he makes in texts.

Brandon Wang — Tech writer and operator, in a widely shared essay.

What they built

A Mac mini running 24/7 at home — chosen specifically for real Chrome browsing, iMessage, and Apple Reminders — with a private Slack workspace as the interface. The agent monitors his texts to detect commitments he’s made and auto-creates calendar events; preps daily meeting summaries; summarizes WhatsApp and Signal group chats; runs 30+ price alerts with reasoning (analyzing hotel photos against his criteria); tracks packages and flights; keeps a freezer inventory from photos; turns recipes into grocery lists; and books restaurants against his calendar.

Why it matters to them

His thesis is that delegation always trades risk for leverage: “the first bits of risk led to a lot more helpfulness.” The agent got useful exactly when he let it into the messy, personal stuff — texts, calendars, errands.

The Octogee version

This is the daily-driver story: a helper in a channel that also includes your spouse, quietly turning conversations into calendar events and errands into done. Octogee’s private-container model is aimed at exactly Brandon’s tradeoff — the leverage of deep access without handing an open-ended tool your 2FA codes on hardware you have to secure yourself.

Sources

$15–35 · per month, cost-engineered from $300+

A content agent that writes in his voice, for $15–35 a month.

Ming Tommy Tang — Bioinformatics researcher (single-cell genomics) and technical writer in Boston.

What they built

A Mac mini setup where his agent drafts X threads, expands blog posts, curates his newsletter, and checks his calendar — with a maintained memory of his writing voice, research area, and audience. “Draft a thread about collider bias in genomics” comes back sounding like him. He cost-engineered it from $300+/month down to $15–35/month by routing hard tasks to premium models and everything else to cheap ones.

Why it matters to them

He wanted “an AI assistant that knew me. Knew my research. Knew my content calendar” — not a generic chatbot, but an amplifier for a domain expert with limited writing time.

The Octogee version

A creator’s helper that accumulates voice and context over time is Octogee’s memory story — without the plumbing. Tommy stitched together five open-source tools to get routing, memory, and messaging working; in Octogee that’s the starting point, and the cost management is someone else’s job.

Sources

70 min · to a DMZ’d, three-agent setup

A CEO’s three isolated agents, set up security-first.

Dirk Paessler — CEO of the Carbon Drawdown Initiative and founder of Paessler AG (the PRTG monitoring company) — someone who thinks about network security for a living.

What they built

A headless Mac mini M4 in a DMZ on his home network, running three agent instances with separate memories: “Number One” (life and work organizer), a carbon-dioxide-removal research agent, and a coding agent — each reachable through its own Telegram bot. The whole setup took about 70 minutes, with roughly 5 minutes of his own involvement.

Why it matters to them

He wrote the guide because the security architecture was the point — isolation between agents, a DMZ’d machine, no shared memory. And he wanted “competent guidance from something that had seen a thousand variations of this setup” rather than hand-rolling terminal config.

The Octogee version

When a network-monitoring founder’s first move is to quarantine his agents in a DMZ, that’s the strongest possible endorsement of Octogee’s default posture: isolated private containers per customer, visible-in-channel actions, no self-managed machine on your home network. The 70-minute setup becomes one click.

Sources

85+ · community use cases catalogued

…and it keeps going.

  • Steve Caldwell runs weekly meal planning through his agent — plan, shopping lists, weather adjustments — saving about an hour a week of household admin.TechRadar’s roundup
  • Albert Moral runs OpenClaw on a Raspberry Pi wired to his WHOOP wearable, acting on recovery and fitness data.TechRadar’s roundup
  • Mike Manzano uses it as an overnight coding orchestrator — “assign tasks before bed, review the output in the morning.”TechRadar’s roundup
  • Shea Johnson, PhD published a free no-code guide to her 24/7 Telegram-driven assistant on a Mac mini M4.Medium
  • Graham Mann catalogued 85+ community use cases — from a voice-controlled kids’ Minecraft server admin to an agent that negotiates apartment-repair quotes over WhatsApp.grahammann.net
  • Y Combinator’s Lightcone Podcast devoted an episode to the trend.YouTube

Why it works

Capacity. Judgment. Joy.

01 · Your agent

Fluid intelligence is free.

Research, writing, code, iteration — capacity on tap, around the clock.

02 · The Playbook

Crystallized intelligence is what matters.

The patterns and checkpoints that usually take years to earn, built in.

03 · Your channels

Collaboration is fun.

An agentic cofounder, grounded in your shared context — ask it, trust the answer, keep moving.

Together, they help you build like you’ve done this before.

The first Playbook

From concept to customer.

You don’t need every skill, a perfect plan, or a complete team before you begin. Take more shots. Find what works. Bring people into something real.

Join us and build something