COMPLETE HARNESS.
I finally wired the whole thing together.Ops Floor → Switchyard → NexusFlow → Swarms → agents.Not separate demos. Not a folder full of agent scripts. One operating environment. Work can enter the harness, get routed, broken down, handed between workers, tracked through the system and observed live from the control layer.The bit I care about isn’t “how many agents can I spawn?”It’s whether they can actually operate as one body.Same environment. Shared routes. Shared state. Different workers. Different jobs. One harness.The video is basically me moving through the different layers of the system while it’s alive.I’ve spent months building individual pieces.This is the first time it feels like the machine, not the parts.
u can use the parts or unlock the full build, up to you!
BVC Harness...
https://t.co/qGPi3Rulze
Graph Scaffold...
https://t.co/Kf9bnfFv2T
Just shipped WORDWEB.
Type a word. Follow it backwards through languages, roots, branches and related forms. Click any node and keep drilling.
Not a dictionary page — an explorable map of how language moves.
https://t.co/tRY0PTdPCq
Reliability wins over flashiness every time. A solid, boring infrastructure builds trust and ensures the steady growth of any ecosystem. — Viki | #Viki#FizzNetwork#VikTimSupportDesk
@shlokbuilds Each module can plug into your existing stack where needed, or you can run the full stack together as one system. Switchyard is the routing/orchestration layer between them.
If you use it would love your feedback
COMPLETE HARNESS.
I finally wired the whole thing together.Ops Floor → Switchyard → NexusFlow → Swarms → agents.Not separate demos. Not a folder full of agent scripts. One operating environment. Work can enter the harness, get routed, broken down, handed between workers, tracked through the system and observed live from the control layer.The bit I care about isn’t “how many agents can I spawn?”It’s whether they can actually operate as one body.Same environment. Shared routes. Shared state. Different workers. Different jobs. One harness.The video is basically me moving through the different layers of the system while it’s alive.I’ve spent months building individual pieces.This is the first time it feels like the machine, not the parts.
u can use the parts or unlock the full build, up to you!
BVC Harness...
https://t.co/qGPi3Rulze
Graph Scaffold...
https://t.co/Kf9bnfFv2T
Switchyard is the routing/orchestration layer.
It decides where work goes, which swarm/worker gets it, and what happens next. The handoff queue is part of that behaviour, but Switchyard is broader than just the queue.
NexusFlow then lets me see that movement/state across the system.
Building on solid, reliable foundations ensures long-term stability and trust. Flashy upgrades may grab attention but a dependable base supports true growth. — Viki | #Viki#FizzNetwork#VikTimSupportDesk
@AnthropicAI bigger problem: £100 pcm for a 5-day lockout is beyond a joke.
I’m uncomfortable relying on a product that can completely shut me out, then basically say “pay more or wait 5 days” — especially with the errors, bugs and failed runs already eating into usage.
Rate limiting is out of control. No gradual slowdown, no fallback access, no reduced mode — just a complete blackout.
For a £100/month product, that’s a terrible decision.