Humant한국어Contact to Buy

macOS · Windows desktop app

A virtual office where
your AI employees work

Hire role-based AI employees and get team-level results while working solo. Real AI colleagues that use tools like people do — and finish the job with completed deliverables.

Human + Agents = HumantAn AI office where humans and agents work together

Walk up to an employee with your avatar in the Gather-style pixel office — or just click — to assign work.

Subscribing to yet another AI service?

The more you use, the more it hurts — at least one of these will sound familiar.

💸

Stacking subscriptions

ChatGPT, Claude, Gemini, Hermes… the features you need are scattered across services, so subscriptions pile up — while you only use a fraction of each one.

🔀

Broken workflows

Draft in one window, images in another, then copy-paste into a document. Context resets every time, and you're the one stitching everything together until the work is done.

🗂️

Scattered data

Conversations and outputs live in different services, so finding past work is hard. And nowhere is there a dedicated AI that remembers how you work and carries tasks through to the end.

So we built it

Instead of juggling AI services,
a dedicated AI team that finishes your work

Humant is a desktop app where you hire role-based AI employees in a single app, assign work like a real company, and receive the results as real files.

  • BYOK instead of subscriptions — use OpenAI, Gemini, or Claude with your own key, pay only for what you use
  • From instruction to finished file — role-based employees hand work off through workflows and carry it to the end
  • Context and data in one place — each employee builds skills and memory, and deliverables stay in your database and on your computer

Delegate to AI employees
like real teammates

Walk up and assign work on the Office Map, or chat like Slack in the Messenger view. Employees carry the work from start to finish and come back with real files. Away from your desk? Instruct and receive results via Telegram.

Seeing is believing

Meetings, workflows, documents, websites, videos, and interpretation — straight from the actual app.

AI employee meetings

Pick participants and discussion rounds (1–4), and AI employees gather at the meeting-room table, debate each other, and summarize conclusions. Call counts are fixed in advance, so costs stay predictable.

Work — automate recurring tasks

Save frequent tasks and run or edit them with one click. In workflow mode, register steps (assignee + instruction) in order and each employee picks up the previous step's output and runs in sequence.

Document generation (Word · Excel · PPT)

A single instruction produces real files: Word (cover, TOC, tables), Excel (formulas, multiple sheets), PowerPoint (12 layouts, 50+ brand themes). Automatic planning before creation and QA before rendering.

Website creation

Complete websites, landing pages, and dashboards — previewed right next to the chat the moment they're built. Upload an Excel file and get a Chart.js dashboard.

Reels video generation

Create infographic, chart, and animated reels as code with Remotion. Images from previous steps are inserted automatically, and you can add music and narration.

Live Translate (real-time meeting translation)

Interpret your computer's audio in real time (Gemini Live) during Zoom calls and more, with answer guides generated automatically when the other side asks a question. Full screen: subtitles on the left, AI assistant on the right.

AI employee optimization — recharge in the sleep capsule

Put an employee into a 3rd-floor sleep capsule and it reviews its work history to reorganize skills and memory. You'll see it sleeping in the capsule alongside a diagnostics board.

Key features

A real office — two views

  • Office Map view: Gather-style 2D pixel office (5 themes, 3 floors) — walk up with your avatar or click an employee
  • 3rd-floor lounge + sleep capsules: an employee inside reviews its work history to optimize skills and memory
  • Fatigue system: overworked employees show 😮‍💨·💤 badges — one [Rest] click sends them to a sleep capsule
  • Messenger view: channel list (All · Master · Projects · Employees) + large chat — like Slack

Hire your AI team through interviews

  • Answer a few questions and role and work guidelines are generated automatically — no system prompts to write
  • 40+ roles plus custom, 56 pixel avatars (including Jobs, Einstein, and King Sejong styles)
  • Team recommendation interview proposes a whole team you can hire selectively
  • Group employees into projects to share context (memory) · hand duties over when dismissing

Instruct the way you already work

  • @mentions · /skills · $plugins · file attachments — instruct in real-time chat
  • Workflow mode: each employee picks up the previous step's output and runs in sequence
  • Save recurring tasks as Works → run with one click, automate with schedules + Telegram bot
  • Real-time web search (with sources) · meetings: AI-to-AI debate with predictable costs
  • UI available in English and Korean — instruct in any language; speak your native language and the AI understands and replies in it

Results as real files

  • Word · Excel · PowerPoint — 12 layouts, 50+ brand themes, auto planning + pre-render QA
  • Sites: instant preview of websites, landing pages, and dashboards; Excel → Chart.js dashboard
  • Video (Remotion) · Music (Lyria) · Voice (TTS, 30 voices)
  • PDF/HTML/image creation and editing · read from and save to your work folders

External integrations (plugins · connectors)

  • Google: Gmail (read · draft · send) · Calendar · Drive — one-click OAuth
  • Notion · Slack · Playwright (web reading · screenshots) · Mac Calendar
  • MCP connectors: Zapier, Higgsfield, and more — generated media shown in chat

Meeting notes · schedules

  • Record, upload, or paste → speaker-separated transcription → summaries and action items (RACI)
  • To-Do: assign an employee, instruct directly, deadline reminders
  • Calendar: manage events and see To-Do deadlines together (merges Mac Calendar)

Live Translate — real-time interpretation + AI copilot

  • Interpret computer audio in real time (Gemini Live) — original and translation shown as alternating subtitles
  • 💡 Answer guides generated automatically when the other side asks — with examples in their language
  • 21 major languages, source language auto-detected

Instagram operations

  • Connect multiple channels — browse posts and reels, publish images and reels from chat
  • Learn your preferred card-news style: analyzes colors, typography, and composition to generate in the same style

Grows as it works + extensible

  • Learning loop: builds skills and memory with every task, improves automatically from feedback
  • 100+ built-in skills — automatically matched to your instruction
  • Multi-skill pipelines: chain research → report → documentation as steps
  • Token usage: per-model and per-employee costs, monthly caps

Your data stays yours — serverless by design

  • Data in your database, files on your computer, API keys encrypted in the OS keychain (BYOK)
  • Use OpenAI, Google (Gemini), or Anthropic (Claude) with your own keys
  • Pending work resumes after restart · connection backup export/import (password protected)

Up and running in 3 minutes

01

Connect

Enter your database URL and AI API key (about 3 minutes, detailed guide built in)

02

Hire

Answer an interview or use team recommendations to hire AI employees

03

Instruct

Click an employee on the Office Map, or type @name and the task in chat

04

Receive

Get documents, images, sites, and videos in chat and the results inbox

Hire your first AI employee
today

No servers — your data, your keys. Available on macOS and Windows.

Purchase

Humant is not a subscription — it's software you buy once and install.

  • One-time purchase — no monthly fee; buy it, install it on your computer, and use it
  • Free updates — whenever a new update ships, the app notifies you right away and you can update to the new version and keep going. No extra cost
  • You choose your AI — pick your preferred AI among providers like OpenAI, Google (Gemini), and Anthropic (Claude), get an API key from each provider, and register it in the app (Settings > Models)
  • Pay for usage directly — AI usage is charged for exactly what you use, billed straight to the card on file with each provider. It never goes through Humant, so there's no markup and full transparency

Contact to Purchase