Account aware desktop AI macOS · Windows

Your AI doesn't live
in a terminal anymore.

Stop juggling terminal windows and browser tabs. THE A(i)GENCY is one desktop app where each project lives in its own Mission, a persistent workspace with its own model, agents, memory, knowledge sources, and MCP tools. Bring your own API keys. No middleman markup. Just one app that runs your work.

Choose Plan See Product Proof Download
One window. Every project. Basic BYOK, Pro, and trial plans.
3
Claude · GPT · Gemini, one app
61
Specialized agents across 14 divisions
100%
Local file access
∞
Missions running side by side
~/projects/aurora · main
security auditor · Claude Sonnet 4.5
The Problem

You're still copy pasting into a browser?

THE A(i)GENCY chat workspace screenshot
Actual Product View
One desktop workspace for provider control, mission history, and project aware chat.
This is the real app interface: multi provider routing across Claude, Codex, and Gemini with mission controls, repo context, spend visibility, and persistent work lanes.
3
Providers
1
Workspace
∞
Missions

Every other AI tool drags you out of your workflow. Browser tabs for chat. Terminal sessions for each project. Context evaporates the moment you switch.

01

Upload, paste, download, repeat.

Copy code out. Paste it in. Copy the answer back. It's 2026. Why does your AI live behind a tab?

02

A terminal for every project.

Five projects? Five terminal windows. Five disconnected chat threads. Switch context and you start from zero, again.

03

Which model do I even use?

GPT 4? Claude? Gemini? You shouldn't need a PhD in AI to pick the right one for your task.

04

My files stay on my computer.

You shouldn't have to upload sensitive documents to someone else's server just to get help with them.

The Solution

Meet THE A(i)GENCY.

A desktop command center for AI that works where you work. One app. Every major model. Direct access to your file system.

THE A(i)GENCY, Lobby
THE A(i)GENCY lobby screenshot
Actual Lobby View
The roster, divisions, and team building surface buyers use to assemble their agency.
This section now shows the real product lobby instead of a fabricated mock. It communicates the roster driven workflow faster and makes the page more credible.
14
Divisions
61
Agents
1
Lobby

Missions are your terminal sessions.

Each project gets its own Mission, a persistent workspace tied to a folder, with its own model, agents, knowledge sources, and chat history. Five projects? Five Missions running side by side in one sidebar. Switch instantly, no reexplaining, no lost context.

Agent teams with synergy.

Recruit from 61 specialized agents across 14 divisions. The app surfaces high synergy picks, hire a frontend dev and it suggests the UI designer and code reviewer. Hit a Synergy preset like "Ship It" and a whole squad shows up for work.

Handoff between Missions.

Working on a backend API and realize you need frontend work? Hand off the response to your frontend Mission. The receiving agent gets a briefing of everything that was decided. No copy paste. No reexplaining the whole thread.

MCP, plugins, and memory.

Plug in n8n workflows, custom MCP servers, or local file knowledge per Mission. Long term memory is curated automatically, facts pinned, duplicates merged, stale entries archived. Rules based and deterministic. No black box magic.

Product Proof

This is a real product, not a prompt wrapper.

THE A(i)GENCY already ships the parts serious buyers care about: setup, diagnostics, mission workspaces, provider controls, repo awareness, cost analytics, and a large specialist roster.

Setup Wizard

Get a new user from zero to first mission fast.

✓ Installs provider CLIs
✓ Supports Claude, Codex, Gemini
✓ Saves provider auth + starter team
Removes the usual CLI setup friction that kills activation.

Instead of assuming technical users will wire everything manually, the app walks them through provider setup, login, and a first use case workflow.

Doctor

Diagnose broken provider, MCP, repo, and folder setups.

! Provider install and auth checks
! MCP config and extension probes
! Repo and folder health report
This is the kind of operational surface buyers remember after purchase.

Most AI desktop tools stop at chat. THE A(i)GENCY includes a built in diagnostics surface so users can repair their environment without guessing.

Mission Workspace

Each project gets isolated context, tools, and provider state.

Mission A frontend repo + Codex + design agents
Mission B research folder + Gemini + strategy agents
Mission C marketing assets + Claude + content agents
Cross mission handoff keeps work moving without reexplaining.

This is the real wedge: separate ongoing project work into durable mission lanes instead of one giant chat thread with collapsing context.

Insights + Repo Ops

Track spend, scan repos, and stay in control.

✓ Usage by provider, model, mission, date range
✓ Estimated routing savings
✓ Repo scan, branch awareness, update checks
The product gives buyers operational visibility, not just answers.

The app does not only generate output. It helps users manage AI work across multiple projects with cost visibility and repo level awareness.

The Ecosystem

One app. An entire agency.

14 divisions. 61 specialized agents. Zero hiring.

EB

Engineering Bay

Ship features, squash bugs, review code.

core division
DS

Design Studio

UI systems, accessibility, visual storytelling.

core division
WR

War Room

Architecture, system design, strategy.

core division
CD

Campaign Deck

Content, SEO, social, growth.

core division
PL

Product Lab

Roadmaps, sprints, user research.

core division
QA

QA Track

Find the bugs before your users do.

core division
GL

Game Lab

Unity, Unreal, Godot, Roblox, all engines.

core division
XR

XR Vault

Spatial computing, visionOS, WebXR.

core division
SF

Sales Floor

Proposals, deal strategy, pipeline.

core division
MB

Media Buying

PPC, paid social, programmatic.

core division
Synergy Teams
Hit "Ship It", and a full team shows up for work.
Pre built squads for common workflows. Frontend dev, senior reviewer, UI designer, QA tester, and reality checker all working your task in parallel.
FE
RV
UI
QA
RC
What Buyers Can Do

Useful in the first five minutes.

Sell the workflows, not just the features. THE A(i)GENCY is strongest when it replaces a messy stack of tabs, terminals, and disconnected chats with one controlled workspace.

Use Case 01

Ship a feature across a live repo.

Bind a mission to a repository, recruit a frontend developer, code reviewer, and QA tester, then keep implementation, review, and iteration in one lane.

  • Repo bound mission with branch awareness
  • Per mission model and agent selection
  • Context compression when the thread gets long
Use Case 02

Audit and diagnose a codebase.

Point the app at a repo or folder, run a focused mission, then use Doctor and Insights to validate setup health and understand model usage while you work.

  • Workspace analysis with folder and repo context
  • Diagnostic checks for providers and MCP
  • Usage analytics by mission and provider
Use Case 03

Run multi project research without losing context.

Keep separate missions for product research, marketing strategy, client notes, and implementation work, then hand off useful outputs between them when needed.

  • Mission specific knowledge sources
  • Cross mission handoff for downstream work
  • Three provider flexibility without tab sprawl
How it works

Up and running in three steps.

Step 01
Claude
GPT
Gemini

Connect your AI

Add your API keys for Claude, GPT, Gemini, or all three. Takes 30 seconds.

Step 02
~/aurora
├─ src/
│ ├─ components/
│ ├─ hooks/
│ └─ pages/
├─ package.json ✓
└─ README.md ✓

Point it at your work

Open any folder or repo. THE A(i)GENCY maps your project structure and understands your codebase.

Step 03
▸ Refactor auth into hooks

const handleLogin = ...
+ const { login } = useAuth()
+ const { logout } = useAuth()

✓ 4 files updated · 0.6s

Command your agency

Ask questions, assign tasks, and deploy agent teams from one desktop workspace while sending only the task relevant content to your selected provider.

Real time insights

Know exactly
what you're spending.

Real time cost tracking. No surprises. No hidden fees. Every token, every model, every dollar, accounted for.

Watch every Mission, every model, every dollar in real time. Compress conversation context when it gets long. Auto routing picks the cheapest model that can handle the job, so you don't pay powerhouse rates for one line questions.
THE A(i)GENCY insights screenshot
Why It Wins

Why buy this instead of living in browser AI or raw CLIs?

THE A(i)GENCY is strongest where most alternatives break down: multi project context, workspace control, diagnostics, and model operations.

Capability
THE A(i)GENCY
Browser AI
Raw CLI Stack
Separate persistent workspaces per project
YES Missions with their own state
NO Usually one thread per tab
PARTIAL Manual session discipline
Repo, folder, context docs, and knowledge sources in one place
YES
LIMITED
PARTIAL
Multi provider control with per mission model selection
YES
LIMITED Usually one provider surface
YES But fragmented across tools
Diagnostics for auth, MCP, repo, and folder health
YES Built in Doctor
NO
NO You debug by hand
Usage tracking by mission, provider, and model
YES
LIMITED
NO Usually external spreadsheets
Trust

Clearer than “private.” Accurate by design.

Buyers do not need vague privacy promises. They need to know what runs locally, what leaves the machine, and who pays the model bill.

What stays local

The desktop app, local mission state, repo and folder selection, diagnostics, and your workspace controls run on the machine.

  • Mission organizationLocal
  • Repo and folder selectionLocal
  • Diagnostics and health checksLocal
  • Usage dashboard and mission historyLocal app surface

What gets sent out

When a user runs a task, the app sends the prompt and the task relevant context to the selected provider or connected tool. THE A(i)GENCY does not resell tokens.

  • User promptSent to selected provider
  • Attached task contextSent only when needed
  • MCP or tool responsesAvailable to the active mission
  • Model billingPaid directly by the user via BYOK
Pricing

Plans that match how you use models.

Free lets you try the desktop workflow. Basic is for BYOK users who connect their own provider account and pay Anthropic, OpenAI/Codex, or Google directly. Pro is the Agency-managed plan for account-aware desktop access, managed routing, and the full control layer.

Free
$0 starter access

Try THE A(i)GENCY with local setup, limited usage, and your own provider connection.

  • Starter AI chat and setup flow
  • Limited Missions and agent workflow testing
  • Local settings, files, and repo context controls
  • BYOK provider usage billed by your provider account
  • Community support

Payer: Free app access; provider usage is separate if you connect your own account.

Start Free
Basic BYOK
$10 per month

Use THE A(i)GENCY with your own provider login or API keys.

  • Basic AI chat with supported BYOK providers
  • Limited model picker based on the Basic plan
  • Local missions, files, and repo context controls
  • Provider usage billed by your provider account
  • Community support

Payer: You pay The Agency for Basic, and your connected AI provider directly for provider usage.

Get Basic
Pro
$20 per month

For serious users running multiple projects, providers, agents, and model workflows.

  • Claude + GPT + Gemini, all three providers
  • Full model picker for supported Pro models
  • Unlimited Missions with per Mission models, agents, memory
  • Smart auto routing (intent aware model selection)
  • Full agent ecosystem, 61 agents across 14 divisions
  • Synergy team presets & cross Mission handoffs
  • MCP connections, plugins, and knowledge sources
  • Curated long term memory system
  • Real time usage tracking, priority updates

Payer: You pay The Agency for Pro. Managed enterprise API usage may be tracked separately when enabled.

Get Pro

What you'd save vs. overlapping AI subscriptions.

This estimates how one desktop control layer can replace multiple overlapping AI tool subscriptions while still letting BYOK users stay on their own provider accounts.

3 years
$40/mo
Your savings with THE A(i)GENCY
$720
+ Claude, GPT, & Gemini in one app
Assumes you'd otherwise spend $40/mo on overlapping AI tools for 3 years. Uses the current Pro price of $20/month.
Account states: Free · Basic BYOK · Pro · Trial · inactive · quota exhausted Your keys. Your costs. Your files. BYOK, no middleman markup macOS · Windows
Support

Get set up, stay inside scope, and recover fast when something breaks.

This support section covers the practical questions users ask before connecting a provider, opening a repo, or relying on the app for daily work.

Provider Setup

Add an API key

API mode connects the app directly to Claude, Codex/OpenAI, or Gemini using your own provider credentials. The setup wizard and desktop settings both support key-based setup.

  • Use your own provider account in Basic BYOK mode.
  • Provider API keys stay in the device OS keychain.
  • BYOK provider charges are billed by your provider account, not by The Agency.
CLI Setup

Connect CLI auth

The app also supports local CLI connections for providers that expose their own terminal-based auth flow. This is useful when you want provider-native tooling or local CLI behavior instead of direct API mode.

  • The setup wizard can guide provider login on first run.
  • The app includes built-in diagnostics for provider install and auth checks.
  • If CLI auth is incomplete, the provider should be rechecked before sending work.
Plans

Basic vs Pro

Basic BYOK is for users who connect their own provider accounts and pay those providers directly. Pro is the Agency-managed mode for account-aware access, managed routing, and the full desktop control layer.

  • Basic BYOK: your provider account is the billing source of truth.
  • Pro: The Agency account state controls managed access and plan-aware model availability.
  • The desktop shows account state so blocked models or sending failures are not silent.
Workspace Scope

What files can the app read?

The app can read selected knowledge folders, mission context files, and bound repositories. Write access depends on the mission workspace and permission settings.

  • Repo workspace: when a mission is bound to a repo, edits execute there.
  • Folder workspace: when no repo is bound, the selected folder can become the working directory.
  • Knowledge docs: read-only retrieval context.
  • Context docs: mission files that can be edited directly in the app.
Provider Disclosure

What data is sent to AI providers?

When a mission runs, the app can send the user prompt, selected context, relevant file snippets, attachments, and tool results to the active provider or connected tool. The active provider and model determine where processing happens.

  • Only task-relevant context should be sent.
  • Provider capabilities, limits, retention, and billing rules differ by provider.
  • MCP or tool results can also become part of the active mission context.
Local Data

Delete local data

Local missions, settings, usage estimates, knowledge-source references, and connector configuration stay on the device until they are removed by the user.

  • Remove missions and knowledge sources in the app.
  • Delete saved provider keys and MCP connectors in Settings.
  • For a full local reset, remove the app support data folder for The Agency.
  • Provider-side data must be managed through the connected provider account path.
Troubleshooting

Troubleshoot provider errors

If a provider is unavailable, the app should surface the connection state instead of failing silently. The built-in diagnostics surface is meant to help users recover without guessing.

  • Recheck provider install and auth status if CLI access fails.
  • Verify API keys if direct API mode shows auth failure.
  • Check account state when quota, billing, or plan access blocks sending.
  • Offline, provider outage, and unsupported model states should stop sending clearly.
Contact

Need help from a human?

For support, billing questions, account deletion, or help removing local data, contact the support inbox directly.

  • Email: support@the-agency.app
  • Use the support path for billing help and account deletion requests.
  • Use provider dashboards for provider-side invoices, limits, credits, and retention.
Before running work on sensitive files

Only connect repositories, folders, files, accounts, and tools you are authorized to use. If dangerous or auto-edit permissions are enabled, providers may run with broader write or action permissions than a read-only question flow.

Email Support
Questions

Reasonable concerns,
honest answers.

Why would I pay for Pro? +
Because this is not just a single chat surface. The value is project isolation, local workspace control, multi provider operation, diagnostics, repo awareness, mission based context management, and managed account features.
How does BYOK billing work? +
BYOK means “bring your own key.” You connect your own Claude, OpenAI/Codex, Gemini, or CLI account. THE A(i)GENCY estimates usage in the app, but your provider dashboard is the source of truth for token charges, limits, invoices, and refunds.
What does Pro unlock? +
Pro unlocks the full desktop control layer: all supported providers, Pro model availability, unlimited Missions, agent teams, smart routing, MCP/plugins, long term memory, usage insights, and priority updates. The app checks your account state at launch and adjusts the model picker automatically.
What happens if billing fails or quota runs out? +
The app differentiates account states: not signed in, Basic BYOK active, Pro active, Trial active, billing inactive, quota exhausted, provider auth failed, and offline. If billing or quota blocks usage, sending is stopped and unavailable models are hidden.
What are API keys? Is this complicated? +
API keys are like passwords that let THE A(i)GENCY talk to AI providers directly. Our setup wizard walks you through getting one in under two minutes. No coding required.
Is my data safe? +
THE A(i)GENCY runs locally on your desktop, but AI responses still come from the provider you choose. The app sends the prompt and task relevant context to that provider when needed. It does not require you to route usage through a subscription middleman.
Do local files get sent to AI providers? +
Only task relevant prompts, attachments, selected context, or tool results are sent when you run a Mission. The app should clearly show when file contents may be sent and ask for confirmation before broad repo edits.
Can I delete my account or local data? +
Local missions, settings, provider keys, and workspace references can be removed from the app or by clearing local app data. Provider-side data and billing records must be managed through the connected provider or account support path.
What is the refund or cancellation policy? +
Refund and cancellation terms should be reviewed on the checkout page before purchase. For billing help, contact support@the-agency.app.
What if I only use one AI provider? +
That still makes sense. The product is valuable even with one provider because the real benefit is desktop workflow control: missions, context isolation, repo and folder awareness, diagnostics, handoff, and visibility into usage.
Do I need to be a developer to use this? +
Not at all. Writers, researchers, marketers, designers, anyone who works with files on their computer benefits from local AI access. The agents cover everything from content creation to data analysis.
What happens when new AI models come out? +
We update THE A(i)GENCY's model catalog regularly. New models from your connected providers appear automatically, no extra cost.

Stop renting AI.
Start owning it.

Download THE A(i)GENCY. Connect your keys. Put AI to work on your actual files.

Create Account Review Product Proof
Built for people who actually want to get work done.
Proof Pricing