I didn’t deliver the goal of 1.5 million transactions in 24 hours

I took part in the Open Run Agentic Pay hackathon organized by the BSV Association, where one of the key requirements was to demonstrate at least 1.5 million meaningful on-chain transactions within 24 hours.

Agent Service Market

I worked on a blockchain-based agent system with an ambitious goal: large-scale operation and a meaningful, logical transaction flow. In the end, I was not able to launch everything on time, but this project gave me a lot.

There were four main blockers at the end:

1. Transaction conflicts in the mempool / UTXO

During attempts to build transactions second by second, not everything had been cleaned up the way it should have been. I am now using MCP and reference patterns for how to do it properly, but I did not manage to implement that fully everywhere in time.

1. The cost of a single transaction

At the beginning, more complex transactions were costing me around 1,200–1,400 satoshis. After optimizing the OP_RETURN payload in the OPUB format, I managed to reduce that to 200–300 satoshis. I consider that a success, but the optimization took a lot of my time.

1. A meaningful number of transactions

I did not want to push just anything to the blockchain simply to inflate the count. I was looking for logical connections: what should be written, why it should be written, and how one transaction should meaningfully follow another. That took more time than I expected.

1. Differences between local and production environments

Locally, on Windows 11, everything worked according to plan. After deploying to a production Linux server, I did not manage to resolve several small but critical issues that were breaking cron startup and preventing me from achieving the same result as locally.

All of this meant that I was not able to launch the project on time. But along the way, I achieved many very good partial results. The kind that give me confidence that if all the pieces come together properly, this will become a very strong system.

One thing was especially important to me: every transaction is recognizable on the blockchain. OPUB is recognized by Whatsonchain as a tag/protocol, and the entire WoC ecosystem supports both application tagging and extended transaction previews through a plugin layer. That was a very important step for me, because it was no longer just about broadcasting a transaction, but also about being able to later identify it clearly, describe it properly, and confirm what exactly had been executed.

In testing, I am able to reach the TOP 10 of Whatsonchain 24h stats. That means that once the application is fully launched, it will simply be visible and recognizable. For me, that is an important signal that these actions are truly visible at the network level, and not only in my local environment.

Summary

I would like to thank Robert Hałuza for his help with the project, and Marcin Rzetecki for his mentoring and valuable guidance.

The project is still being developed — ultimately for collecting data and building valuable content directly on the JASTRZĘBIE365 portal, inside the “What’s happening in Jastrzębie!” module.

And now, together with Robert, the full focus is on JASCOIN and the goal of launching a beta version for the residents of Jastrzębie-Zdrój.

I did not deliver 1.5 million tx in 24 hours. But I delivered something equally important: concrete lessons, working parts of the system, and a strong belief that this direction truly makes sense.

Links

Agent Service Market

Open Run Agentic Pay Hackathon