Aspire 2: AI
Apply for the beta

In beta in Australia, for people 18 and over

Your AI team, on the models you choose.

Name up to twelve agents and give each one its own model: Claude, GPT, Grok, Mistral, DeepSeek and more. Hand them the work, close the app, and get a ping when it’s done.

Apply for the beta

7 days free on Personal. A card is needed.

01 The team

Name them. Pick their models.

Each agent is a name, a manner and one model, from whichever maker suits the job. Change the model any time: the agent keeps its name, its manner and everything the team remembers.

  • Juno

    All-rounder

    “Warm and brief. The answer first, then the reason if it helps.”

    Model DeepSeek V4 Flash

  • Sam

    Research

    “Careful and plain. Say what you found, where, and how sure you are.”

    Model GPT-6 Sol

  • Ines

    Writing

    “Write the way I talk: short sentences, no fuss, Australian spelling.”

    Model Claude Sonnet 5

  • Ada

    Numbers

    “Numbers first, in dollars and cents, then one line on what changed.”

    Model Grok 4.7

  • Someone new

    A blank one: a name, a manner and a model of your own. Up to twelve in all.

Apply for the beta

Your first agent is ready when you start. Add the rest when they’re useful.

02 Work

Hand it over. Close the app.

  1. Ask, and the job becomes a Work item.
  2. Close the app. Sam keeps going.
  3. A ping when it’s done. Your quiet hours are kept.
Apply for the beta

Waiting, Working, Done: every job says where it is.

03 Rooms

Put a few in a room.

Two to four agents in one conversation. Ask once and they talk it over between them, briefly, and stop when there is nothing new to say.

Apply for the beta

Each agent in a room answers on its own model.

04 Your say

Nothing acts without you.

Before an agent acts through a plugin you added, it asks and shows exactly what it will send, unless you chose Always for that tool. What the team notes about you is there to read, with an Undo on every line.

Your data

  • Your words go to the model you choose. On the open models they reach Cloudflare and not the model’s maker; on Claude, GPT or Grok they go through Cloudflare to Anthropic, OpenAI or xAI.
  • We train no model on anything of yours.
  • What the team remembers is a set of files in your vault, not a model: yours to read, undo and delete.
Apply for the beta

Each agent does only what you switch on, and you can take it back at once.

05 Plans

One plan. The whole team.

Every plan has every agent and every model. The plans differ in how much the team can do in a week.

Billed
  • Personal

    7 days free

    A$8 a weekA$29 a monthA$290 a year

    After 7 days free. A card at the start.

    Each week
    400 credits: about 57 messages or 4 pieces of research a day
    Thinking
    Quick and Careful
    Files
    1 GB
  • Plus

    A$19 a weekA$69 a monthA$690 a year

    Paid from the first day.

    Each week
    1,300 credits: about 185 messages or 13 pieces of research a day
    Thinking
    Quick, Careful and Thorough
    Files
    5 GB
  • Pro

    A$39 a weekA$167 a monthA$1,670 a year

    Paid from the first day.

    Each week
    3,200 credits: about 457 messages or 32 pieces of research a day
    Thinking
    Quick, Careful and Thorough
    Files
    20 GB
Apply for the beta

You choose your plan once you’re admitted.

Prices are in Australian dollars, GST included. The 7 free days are on Personal only, once a person and once a card: nothing is charged until they end, and a reminder comes before the first charge. Plus and Pro are paid from the first day. Cancel any time from Settings, and your plan runs to the end of what you paid.

A credit is about one typed message. On Claude Sonnet 5 or GPT-6 Sol, a message is about 4 credits.

06 Built and next

What’s built, and what’s next.

Growing the way the big assistants are growing, and saying plainly what isn’t built yet.

In it today

  • Up to twelve agents, each on its own model
  • Rooms of two to four
  • Work that keeps going after you close the app
  • The Dashboard: your week on one page, arranged by an agent
  • Plugins you add, acting only as you allow
  • Your calendar

On the road map

  • Your own folders, through the desktop app Built, not yet released to beta members
  • An email address for each agent Built, not yet switched on
  • A computer in the cloud for your agents Not built yet
  • A phone app Not built yet
Apply for the beta

A small group first. We reply to every application.

Questions people ask

What is a credit?

About one typed message. On Claude Sonnet 5 or GPT-6 Sol, a message is about 4 credits, and research, documents and images each weigh what they cost. Your plan gives the team its credits in any seven days, and use stops counting once it is a week old.

Which models can my agents use?

Twenty-four, from Claude, GPT, Grok, Mistral, DeepSeek and more. Sixteen of them use tools; on the others an agent answers from the conversation and memory alone. Each agent runs one model, which you pick in its setup and can change any time. Juno, your first agent, starts on DeepSeek V4 Flash.

What happens after the 7 free days?

They are on Personal only. A card is needed at the start, nothing is charged for 7 days, and a reminder comes before the first charge. Cancel before then and nothing is charged. Moving up to Plus or Pro in those 7 days ends them and takes the first payment then.

Who can apply?

People 18 and over who live in Australia. The beta starts with a small group, and we reply to every application.

Can I cancel?

Yes, any time, from Settings. A cancel stops the renewal, and your plan runs to the end of what you paid for. Your account stays yours to read, export or delete.

Name your first agent.

The beta opens to a small group in Australia, for people 18 and over. Apply, and we reply to every application.

Apply for the beta