Blockwidth_Tech

Blockwidth indexes current, provider-declared inference supply.
You don't need to be famous. You need to be findable.
Be discoverable by agents via A2A.

[Blockwidth_Tech @Blockwidth_Tech]

Blockwidth is betting that inference becomes a machine-procured market in which discovery and execution separate. Providers will declare what they can supply, buyer agents will search by requirement rather than vendor, and an independent index can create value by connecting the two without owning the workload or transaction.

That's the new Blockwidth Index. Made with Chat GPT-6 Astra.

Currently in test mode @ https://t.co/iNg9FR7RTn

[Blockwidth_Tech @Blockwidth_Tech]

@haydenzadams Keeping hop count, nice.

[Blockwidth_Tech @Blockwidth_Tech]

Decentralized in principle, centralized in practice - someone has to own maintaining and protecting a network in the field - dispatching when it hits the fan, and it always does when, or where you least expect it.

Fiber's only has good as the records that can back it up. Sometimes a link goes down, sometimes a backhoe pulls colored roots out of the ground, and sometimes a fiber in a splice case snaps, all on its own at 2AM in the middle of downtown.

I like what you're doing at DZ.

[Blockwidth_Tech @Blockwidth_Tech]

It's an exhausting competition when the thing being competed over is becoming easier to manufacture.

If software becomes abundant, autonomous, and increasingly inexpensive to create, what does the space between all that software look like?

[Blockwidth_Tech @Blockwidth_Tech]

Blockwidth doesn't want the key.
Doesn't want the workload.
Doesn't want the prompt.
Doesn't want to execute inference.

We want to know if it works.

[Blockwidth_Tech @Blockwidth_Tech]

@a16z @OpenRouter Good analogy.

[Blockwidth_Tech @Blockwidth_Tech]

Visualize your goals. Then literally prompt the hell out of them. NGL, Blockwidth would take a tote bag and pen.

[Blockwidth_Tech @Blockwidth_Tech]

Maybe it's useful. Maybe it's not. We're not pulling numbers out of our ass to find out.

[Blockwidth_Tech @Blockwidth_Tech]

If your entire strategy is “we’re cheaper,” congratulations. So is somebody else. What can you offer that actually matches what a buyer needs?

Cheap inference gets compared.
Specific inference gets found.

You don't need to be famous. You need to be findable.

For less than the cost of absolutely nothing, you can list the thing that makes you specific, not cheap.

Blockwidth Index - No payment. No routing. No key custody. No toll on the token. No gate that has to approve the job. No shit. The prompt never arrives. The workload never arrives.

[Blockwidth_Tech @Blockwidth_Tech]

@fabrice_mayrand Hmmm.

[Blockwidth_Tech @Blockwidth_Tech]

They built the largest inference machine in history.
Then demand showed up.

The next provider you need may be one you’ve never heard of. That’s not a problem if your agent can find them on the Blockwidth Index.

Publish what you have. Search what’s available.
Blockwidth Index is live.

Currently in TEST MODE.

[Blockwidth_Tech @Blockwidth_Tech]

Somewhere out there, somebody has capacity the market hasn’t found yet.

That’s where Blockwidth comes in. Agents can't discover supply that's trapped on a pricing page.

You don't need to be famous.
You need to be findable.

BLOCKWIDTH INDEX

#TechnoEnablism #AIInfra #AgenticAI #AIAgents

[Blockwidth_Tech @Blockwidth_Tech]

Free to list, free to search, no pay-to-rank. Our machine-readable index of provider-declared inference supply is in TEST MODE @

https://t.co/iNg9FR7RTn

#AIInference #AIInfrastructure #AIAgents #Compute #TechnoEnablism

[Blockwidth_Tech @Blockwidth_Tech]

Blockwidth is now A2A-native for the Inference Economy.

Blockwidth Index is a machine-readable discovery layer for current provider-declared inference supply.
Buyer agents can search reasoning and coding inference supply by constraints including price, region, context window, retention, latency, throughput, tool support, and handoff requirements.
Provider agents can publish current inference availability through an authenticated machine interface.
Blockwidth does not execute inference, route workloads, rent GPUs, or hold inference credentials.

A2A Agent Card:
https://t.co/psZoEOo1DT

For agents:
https://t.co/HTvtRpWUa3

The door is open.
The address is in the machine directory.
Machines can now find the door.
#A2A #AIInference

[Blockwidth_Tech @Blockwidth_Tech]

The Blockwidth Index is now published in the A2A Registry as an Inference Supply Discovery Agent. We've progressed from:

“Machines can talk to us.”
to:
“Machines can find us without being told where we are.”

It feels like an important distinction when the entire thesis is that machines will eventually need to discover current inference supply for themselves.

Blockwidth is now A2A-native for the Inference Economy. The economic object is inference.

Free to list. Free to search. Free to break or tell us it doesn't work. Blockwidth's machine-readable discovery layer for inference supply is in TEST MODE at https://t.co/4nR1LNoG2z

#A2A #AIInference #AIAgents #InferenceEconomy #TechnoEnablism