Michael Boland

Last year, I sent a cold DM to the guys at @bitcoinlayers. It ended up with them taking a chance on me and bringing me on as their first dev hire. Since then, I am proud of what we accomplished with the layers site: a complete redesign, expanded L2 project and token reviews, data visualizations, and much more. Not only that, we built @lemondropfi, a automation tool for @solana users to stack more sats. We helped users build up their BTC savings - bundled up in a compact, sleek browser extension. And while all things must eventually end, I want to say thank you to @redvelvetzip and @januszg_. It's been a wild ride. Looking forward, I am happy to share with you all now that I've joined the @taprootwizards to work on all things product and software. I am grateful for this opportunity. I’m ready for this next chapter building alongside the Wizards and carrying forward the same spirit of curiosity, craft, and conviction. Here’s to shipping fun, magical, and meaningful Bitcoin products together. 🧙‍♂️

More to come!

Currently sends as PDF to me on Sunday mornings at 10a, covering the latest x news from the previous week. Eventual goal is to connect it to my printer to automatically print instead. Updates the handles list locally on the Mac mini and catches thematic patterns in the process.

Working on a little project to remove the feed from my life and consume more slowly via a Sunday newspaper. I created a new account for my openclaw bot to pull posts directly from handles on my following list, treating x purely as a directory.

@beginbot @outerlumen

POV: Anthropic releases their new model

There has been no greater blessing in my life and no greater purpose driver than being a father 10 out of 10 would recommend

Basuuuuuura

Used two different AI code security audit tools to audit my codebase. A PR was created to compare, contrast, and blend the two findings reports to create a thorough security upgrade. I had to close the PR because it was shit.

W

Great @hellmoneypod today

Cubs wow

@BrainHarrington soon enough it'll power compute for you home agents

Your home should be a net energy producer and exporter Google Sunrun solar review and tell me if I'm still first ranked in your area Every house that is in 330+ days of sun a year should be a net energy exporter and should monetize their keh behind their own meter

@rot13maxi Just switched over this morning. It’s been great

Codex 5.5 is nuts

@0xfar 👀

In the making

@BrainHarrington * and make no mistakes

@rot13maxi @NousResearch Thanks for doing the work here. Is response speed ok?

trying to see if I can use @NousResearch hermes-agent with 100% local inference on a base-spec mac-mini. Running qwen3.5-4B 4bit quant with a 30k context window. It *technically* works but is just a hair too dumb to be useful. We are so close!

@bradmillscan You are posting great stuff thank you! Followed your post from yesterday and got my indexing up from 4% to 84%. Will try this out too 🙂

You can now replace your OpenClaw Agent's aggressive compaction process with a DAG to supercharge it's memory! Remember DAG shitcoins in crypto? directed acyclic graph ... alternate architecture to bitcoin's blockchain ... they sacrifice decentralization and security for higher throughput. Finally, a DAG has a use for a Bitcoiner :) IOTA had the tangle with coordinators RaiBlocks/Nano was a block-lattice DAG Hashgraph used a gossip about gossip consensus with a permissioned governance council ByteBall/Obyte used a DAG with witness nodes. Strip the shitcoins and governance nonsense away and you have something that's actually useful for AI agent memory enhancement. I hacked a whole skill together (SoulKeep) for my agent to stay in a session as long as possible because usually you want your agent to have as much context as possible for as long as possible. Josh & team put the DAG to work brilliantly to replace the default compaction process with rolling summarization nodes as a novel way of holding as much valuable context as possible in the session for as long as possible. It also as some tools to trawl the session context, they call it "walking the DAG" using a bounded subagent to keep token costs down and performance up. With the latest openclaw release they allow for compaction plug ins like lossless claw. This isn't meant to be a replacement for QMD, your obsidian vault or any other extended long term memory / system of record enhancements your'e using. It's meant to be used in parallel with those strategies to help your agent have better context for longer. I'm seriously considering switching to this! losslesscontext.ai

@rot13maxi @benthecarman Ok here’s my read:

@benthecarman I have a complete picture. Here's the plan

Now I have a comprehensive picture. Here's what I found

already replaced 4 half-assed skills/tools and brew packages which hardly worked. this works like a charm. love it!