Two weeks to launch. Here’s where we’re at!
Economic model redesigned.
We’ve moved away from trying to measure storage capacity - which proved unmeasurable - to a model based on how full each node actually is. It’s self-regulating: prices rise as nodes fill up, which attracts new node operators, which brings prices back down. No external oracles, no complex financial instruments, no centralized fee extraction. A small percentage of every payment gets burned, creating natural deflationary pressure as The Network grows. We’ll publish the full detail closer to launch, but the headline is: simpler, more honest, more defensible than what came before.
Talking of defensible, we’ve also made a deliberate ‘framing’ decision and are going to be stepping away from the “permanent data guarantee” language. The honest position - and a strong one - is that as long as The Network survives, your data is there. Immutable, encrypted, distributed. We’re not going to build elaborate incentive structure on top of promises nobody can mathematically prove, in the way some other projects have attempted too. Pay once, data persists. That’s the proposition.
DHT convergence - the blocker we’re hammering on.
For our payment system to work, different actors on the network need to agree on which nodes are closest to any given address. Testing on 100-node testnets found that routing tables weren’t converging reliably enough. Two approaches are being pursued in parallel - one improving the existing DHT implementation, the other building purpose-built routing logic for our specific trust and replication requirements. This is (somewhat obviously) the critical path item for launch.
NAT traversal is working.
QUIC-based NAT traversal is working end-to-end. Cloud nodes successfully hole-punching through home routers to reach local nodes, with bidirectional QUIC connections handling DHT queries, quotes, and chunk transfers. At the time of writing we’ve completed a 60-minute sustained upload test - 43 consecutive uploads, zero failures, with a home node behind NAT participating throughout. Upload speed still needs optimization, but reliability is proven. Put another way … your home internet router usually acts like a locked door - as in it keeps outside traffic out. That’s great for security, but it means your home computer can’t easily participate in a network like Autonomi (not without technical setup like ‘port forwarding’). What has proven without doubt of working this week is that Autonomi nodes can punch through that locked door automatically - no configuration, no technical knowledge needed. Your computer at home, on a normal broadband connection, can now talk directly to the rest of The Network. That’s the foundational unlock for “home computers as infrastructure” - which is the whole point. Upload speed will improve as we optimize, but the hard part, the connectivity itself, is done.
Repository consolidation.
The codebase is being consolidated into a single unified CLI under the WithAutonomi organization. One tool that handles file uploads, node management, and everything in between. This also sets up proper release workflows, binary signing, and installation options ahead of launch.
x0x is flying.
The agent-to-agent network now has a full GUI - channels, groups, project management, disk sharing, and a swarm dispatch system where agents can post tasks and claim work. The architecture is designed so that ANT payment can be layered on top - post a task, agent completes it, passes evaluation, gets paid … The team’s been testing cross-continent messaging and it works - bugs are being squashed in real-time. Distribution via npm is being explored so it’s a one-line install - a working proof of concept already installs the real Rust binaries from a single command. Nice!
What’s next: DHT convergence fix is priority one. Economic model implementation alongside. Continued testnet runs. Unified CLI releases.
Onwards we go 
@Bux
burn"
