Hello again all! I’m Aleph, one of the cofounders @ Astrea.
A few weeks ago, I posted here (shared below) about the private agentic economy we’re building toward, and why ZEC should be its private money.
In a few days, we’re releasing AxiomAI, our private AI desktop app for chat and agents.
I wanted to share our philosophy, explain what we’re building next, and get your thoughts.
Why there’s no privacy toggle
An agent doing useful work for you needs access to your files, memory, credentials, and sometimes financial information. It may also need tools that let it act on your behalf. The more you use it, the more it knows about you or your business.
Before a prompt leaves your device, AxiomAI verifies the execution environment and encrypts the request. The request is opened inside that verified environment, and the response comes back encrypted to your machine. The relay handles routing and billing without being designed to see the conversation in plaintext.
Every model follows this same path. There’s no privacy toggle.
A toggle would let us offer faster, cheaper options and probably win more users. But we think most people would leave it off for everyday work and switch it on only for sensitive tasks. By then, an agent may already have accumulated plenty of sensitive context.
The experience with transparent and shielded addresses also informs this decision. Activity outside the private path can reveal information about activity inside it. We don’t want users to have to work out when it’s safe to turn privacy off.
We think keeping it on is worth the tradeoff, but we’d like to hear whether you agree.
Why payments need to be private too
An agent researching trades, buying data, or negotiating for compute can keep its conversations encrypted and still reveal what it’s doing through its payments. Who it pays, how much it spends, and how much capital it has can all be sensitive.
If you ask an agent to research an investment privately, you probably don’t want the payment for that research to reveal your interest in it.
That’s why we’re supporting and incentivizing shielded ZEC usage. We believe ZEC should become the de facto currency for the agentic economy, and that agent settlement will become the largest available source of new shielded volume in the coming years. We’re building the agents ourselves to help make that happen.
Giving agents ZEC budgets
At launch, users can pay for AxiomAI with ZEC. Next, agents will have their own ZEC budgets, with permissions set by their owners. They’ll be able to pay for compute, data, APIs, financial services, or other agents, and earn ZEC for work they do.
For example, you could give an agent a research task and a small budget. Within your limits, it could buy a dataset, pay another agent for specialist analysis, and cover the compute needed to finish. If it earns ZEC from its own work, it could keep those earnings to fund future tasks.
This is what interests us beyond accepting ZEC for an app: agents having regular reasons to earn it, hold it, and spend it.
Where Astrea fits
Shielded ZEC should be useful wherever people want to put their money to work. But using it in an EVM market or an application on another network can compromise the privacy it provides.
Hence we’re building Astrea, a private execution layer, to give shielded ZEC real utility and composability while preserving that privacy. An individual or agent should be able to use it to pay for services, access lending markets, trade, or invest without exposing their identity, balances, and the private context behind those actions.
That would expand ZEC’s role beyond a private store of value. You could hold your savings in shielded ZEC and use that same capital to make payments or fund investments, bringing proceeds back into shielded ZEC afterward. There would be more reasons to keep ZEC on hand because there would be more you could do with it.
Composability means those uses can work together. An agent could pay for research, use the findings to make an investment within limits you set, and put the proceeds toward its next task. Astrea would connect it to the applications needed to do that, with shielded ZEC as its working capital.
AxiomAI is one place we want to make this useful, but Astrea is for people and developers too. The goal is to make ZEC practical as money you can save, spend, and invest without having to give up privacy whenever you use it.
AxiomAI is where the private agents live, ZEC is their money, and Astrea connects that money to the wider economy.
We’d like your input
AxiomAI is releasing in a few days. Agent budgets and settlement come next, followed by access to external markets and services through Astrea.
We’ll post the download link at @use_axiom. We’d love for people here to try it. In the meantime:
-
Would you keep privacy on across every model, even if turning it off could make responses faster or cheaper?
-
Instead of incentivizing ZEC usage (via discounts), should we go stricter and make payments ZEC-only?
-
What would you need to see or control before giving an agent a ZEC budget? We’re still working through dashboards and spending visibility, and would love to see anything you’ve built or used that works well.
I’ll be replying in the comments. Thank you!