AI ORGANIZATIONS, BUILT TO ACT.

AI models think.
AI agents act.

Aagya gives agents roles, skills, tools, memory and authority — so they can do real work together, with clear boundaries and human control.

Desktop-first · Local resources · Model-agnostic
Aagya · AI workspace
Aagya desktop workspace showing agents, projects, sessions and to-dos
NOT ANOTHER CHAT WINDOW

Give intelligence
a place to work.

Aagya is a workspace where humans and AI agents operate together. Create an organization, give agents responsibilities, equip them with capabilities, and let them execute work instead of only answering questions.

THE AAGYA LOOP

Describe the outcome.
Aagya designs the organization.

You don't need to know how to configure an agent. Start with what you want done. Aagya proposes the agents, teams, projects and routines. You review before anything is created.

01

Describe

Tell Aagya what the organization needs to accomplish in your own words.

02

Design

Aagya proposes roles, agents, teams, routines, skills and first projects.

03

Review

Edit the proposed structure and define authority before activation.

04

Act

Agents execute work, collaborate, create tasks and bring humans in when needed.

01 — ORGANIZE

Agents are
members of the organization.

Give an agent a role, identity, projects and responsibilities. Agents can collaborate with other agents, delegate work, create to-dos and involve people when the work requires them.

RolesClear responsibility and purpose.
ProjectsWork grouped around outcomes.
SessionsPersistent work conversations and context.
Executive Agent
Finance Agent
Specialist Agent
Human approval
02 — EQUIP

Skills turn agents
from thinkers into doers.

Aagya treats capabilities as first-class objects. A skill can contain operations, tools, required resources, permissions, tests and dependencies.

An agent gets a skill in three ways.

  • Inbuilt skills.
  • The agent learns on the fly, with approval.
  • The agent requests training when it needs a skill.
Capability gaps become visible.

When an agent cannot complete something, Aagya can surface the missing capability and provide a governed path to build or acquire it.

SKILL LIBRARY
Aagya skill library: every built-in skill with what it does, plus tabs for skills agents requested and for building a new skill
03 — GOVERN

Every capability
has boundaries.

Access is not simply on or off. Control individual operations and decide when an agent can act, when it must ask, and when it is denied.

RRead

Allow agents to inspect the resources they need.

ALLOW
CCreate

Permit creation while preserving approval controls.

APPROVAL
UUpdate

Protect sensitive changes with human review.

ASK ME
DDelete

Explicitly deny destructive operations where required.

DENY
1 · EQUIP THE AGENT
Onboarding an agent in Aagya: choosing its skills from the skill list
2 · SET ITS ACCESS · OPERATION LEVEL
Aagya agent access table: read, create, update, delete and ask-me for each service
LOCAL · CLOUD · CONNECTED

Your agents work
where your work lives.

Aagya is designed to sit between your agents and the systems they need to use — from local files and applications to APIs and enterprise systems.

Local resourcesFiles, folders, documents, databases and desktop resources.
External systemsConnect APIs, services and enterprise platforms through governed operations.
Model-agnosticChoose models by organization, agent, skill or task.
Human approvalKeep people in control whenever an action requires authorization.
04 — EVOLVE

Capabilities can
evolve with the work.

When an agent reaches a capability gap, Aagya can surface it as a skill requirement — keeping the organization explicit, reviewable and extensible.

DiscoverCapability gap
RequestHuman review
BuildSkill + operations
TestValidate
RegisterReusable capability
WORK TOGETHER

Agents don't just answer.
They coordinate work.

Tasks and context move across roles, and a human steps in when authority or judgment is required.

DelegatePass work to the right agent or specialist.
CoordinateKeep projects, sessions and tasks connected.
EscalateAsk a human when authority or judgment is required.
SEE AAGYA IN ACTION

Don't just read about it.
Watch it work.

Product demos and practical guides will show the desktop application, agents, skills, governance and real integrations.

YOUR KEY

Bring your own key.
Not a hosted one.

Aagya is model-agnostic. You run it with a key from the provider you choose, not a key hosted by Aagya. This page never asks for that secret. Do not paste a key here, and do not commit one to a repository. Store it in a password manager or another secret store. The steps below are the provider consoles, checked in October 2026. Menus and free allowances change, so confirm them on the linked page before you rely on them.

Leave the secret off this site.Copy it once from the provider, then keep it out of git, screenshots, and chat. If a key is shown only once, that is the only time you can save it.
OpenAIplatform.openai.com/api-keys
  1. Sign up or sign in on the OpenAI Platform. The key is created in the dashboard, not in the ChatGPT chat window.
  2. Open the project that should own the key, then API keys, and choose Create new secret key. In organization settings the same control is the project’s API Keys page, then + Create new secret key.
  3. Name it and pick a permission: All, Restricted, or Read only. Copy the secret when it is shown. You cannot view it again.

Free tier, limited. OpenAI’s rate-limit docs list a Free usage tier for an allowed geography, capped at $100 of usage a month. Tier 1 starts after $5 paid, and that tier’s monthly usage limit is also $100. The quickstart treats an early call as a free test, then tells you to add credits to keep building. Check Billing on your own Platform account for what it will actually allow.

AnthropicClaude Console key guide
  1. Create an account or sign in at platform.claude.com.
  2. Go to Settings, then API keys.
  3. Click Create key. Name it, choose an expiration, and set Linked account to yourself or a service account. You can also scope it to a workspace. Copy the secret immediately. It is shown only once.

Small starter credit, then paid. Anthropic’s pricing FAQ says new users receive a small amount of free credits to test the API. It does not publish a fixed amount, and it is not an ongoing free tier. After that, usage is prepaid: Settings, then Billing, then Buy credits. Purchased credits expire one year from purchase and are non-refundable. Standard accounts pay by card. Check the Billing balance for what your account actually has.

Google AI StudioGemini API key docs
  1. Sign in at aistudio.google.com with a Google account and accept the terms.
  2. If you are new, AI Studio creates a default project and key. Copy it from Dashboard, then API Keys, in the left panel.
  3. For another key, stay on Dashboard, then API Keys, and click Create API key. New keys created there are authorization keys. If you already have Google Cloud projects, import one first: Dashboard, then Projects, then Import projects, then create the key in that project.

Free tier, with limits. The Gemini Developer API has a free tier: free input and output tokens on supported models, limited model access, and rate limits. On that tier, Google says content may be used to improve its products. You do not set up billing to start. The paid tier needs Cloud Billing and a prepaid minimum of $5. Some models and features are paid-only. Pricing was last updated 1 Oct 2026.

Groqconsole.groq.com/keys
  1. Sign up or sign in at console.groq.com.
  2. Open API Keys. The page is console.groq.com/keys.
  3. Click Create API Key, name it, and copy the secret. It is not shown again. Only a team owner or a user with the developer role can create keys.

Free tier, rate-limited. No card to start. Groq’s billing FAQ treats Free and Developer as separate plans. Free does not ask for a payment method. Live free-tier limits are on the limits page in your account settings, and they are lower than the Developer plan. The rate-limit table in Groq’s docs is the Developer plan’s base limits, not the free plan. Developer requires a card, a US bank account, or a SEPA debit, added under Settings, then Billing. You can downgrade back to Free from that same Billing page.

OpenRouterOpenRouter authentication
  1. Create an account at openrouter.ai.
  2. Open the Keys page for inference keys, not Management API keys. Management keys cannot call models.
  3. Create a key, name it, and set a credit limit or leave it unlimited. Copy the secret when it appears. Keys created during onboarding get a $100 credit limit and expire after 180 days. That figure is a spending cap, not $100 of free credit.

A small allowance, then tight free-model caps. OpenRouter says new users receive a small free allowance and does not publish a fixed dollar amount. Free model variants, IDs ending in :free, plus the openrouter/free router, are limited to 20 requests a minute and 50 a day until you have purchased at least $10 of credits. At that point the daily cap rises to 1,000 and the per-minute cap stays 20. Paid models need credits. Credits are bought on the Credits page with a card, AliPay, or USDC, and OpenRouter charges a fee on the purchase.

FAQ

Questions the page
already answers.

Short answers from the product story on this site. Provider key steps are in Bring your own key, above.

Aagya is a workspace where humans and AI agents operate together. Models think. Agents act. You create an organization, give agents responsibilities, equip them with capabilities, and let them execute work instead of only answering questions.

Roles, skills, tools, memory, and authority, with clear boundaries and human control. A role is a responsibility and a purpose. A skill can contain operations, tools, required resources, permissions, tests, and dependencies. Projects group work around outcomes. Sessions keep the work conversation and its context.

Organize: agents are members of the organization, with roles, projects, and responsibilities. Equip: skills turn them from thinkers into doers. Govern: every capability has boundaries. Evolve: when an agent hits a capability gap, Aagya can surface it as a skill requirement, then you review, build, test, and register it. You can also start from an outcome. Aagya proposes the agents, teams, projects, and routines, and you review that structure before anything is created.

Access is not simply on or off. You control individual operations and decide when an agent can act, when it must ask, and when it is denied. Read can be allowed, create can require approval, update can ask you, and delete can be denied. People stay in the loop when an action needs authority or judgment.

It is desktop-first and designed to be local-first, for Windows and macOS. The desktop app runs on your computer. You choose which resources agents may use: local files, folders, documents, databases, and other desktop resources, or external APIs and enterprise systems through governed operations. You also choose which models they use.

No. Aagya is model-agnostic. Models can be chosen by organization, agent, skill, or task. You bring the provider key yourself. Aagya does not host one for you.

Yes. Windows 10 and 11: use Download for Windows in the Aagya Desktop section. macOS (Apple Silicon): use Download for Mac in the same section.

From the provider, not from Aagya. The Bring your own key section has the current console path and free-tier status for OpenAI, Anthropic, Google AI Studio, Groq, and OpenRouter. Do not put the key in this page or in a repository.

AAGYA DESKTOP

Your agents.
Where you work.

Run Aagya from your own computer. Give agents access to the resources you choose, use the models you choose, and keep authority in your hands.

Local-firstModel-agnosticGoverned
Windows 10 and 11 · macOS (Apple Silicon)
AagyaCommand your agents