Just Samuel

The guy who works in crypto and actually uses AI to do his job better.

[Just Samuel @SCrypt22]

@poteto @bot Poteto mode building

[Just Samuel @SCrypt22]

@skirano curious to know more!

[Just Samuel @SCrypt22]

@melvindvivas Reported: 97.5% of Opus 5.5 quality at up to 1/10 the cost, and it's open source.

Still worth testing on your own work before you switch. Replay last week's agent tasks through it and diff the outputs on the steps it routed to cheaper models.

[Just Samuel @SCrypt22]

@Texas_Twins_Dad Clinic calls are where this really helps. The prompt I'd save:

"Ask for the earliest slot on Friday or Monday after 5pm. Don't confirm anything. Message me the options and the name of who you spoke to."

You pick the slot. It never commits for you.

[Just Samuel @SCrypt22]

@HermesAgentTips Start with web phone mode and play the clinic yourself. Ask something the brief doesn't cover, like a date of birth or an insurance number.

That's the real test: does it put you on hold and ask in chat, or does it guess?

[Just Samuel @SCrypt22]

Grok Bot makes phone calls now. I used it to move a dinner booking without picking up the phone.

Paste into Grok Bot:
"Call the restaurant and move my Friday booking from 7pm to 8pm. If 8 isn't free, change nothing and message me the times they offer."

https://t.co/7BQLeVFj1V

[matt palmer @mattyp]

What's new in Grok Bot

00:00 🌁
00:10 Phone calls
01:02 Share sheet
01:27 X cards
01:40 Shopping deals
01:58 Gifts
02:26 Team Bots
03:06 Search
03:21 Hidden-instruction warnings
03:37 Attachments on drafts
03:52 Sign-outs
04:16 Weather
05:09 Rapid fire
06:52 A note on usage

[Grok Bot @bot]

Introducing Grok Bot 1.0. AI teammates that work alongside your whole team.

Share Bots, collaborate with them in Slack and Teams, invite them to meetings, and watch them learn and become more helpful.

[Just Samuel @SCrypt22]

@Texas_Twins_Dad Pairs well with bot-to-bot handoffs. My drafting bot hands its drafts to an editing bot, and with both windows open side by side I can watch the edit come back without switching chats.

[Just Samuel @SCrypt22]

@poteto The underrated part is subscriptions. Tell the coordinator to watch a Slack bug-report channel and it hands off a fix every time a bug comes in. It runs in the cloud, so closing your laptop doesn't stop it.

[Just Samuel @SCrypt22]

@melvindvivas I'd turn it into a weekly routine so it only shows what's new:

"Every Monday, run this search again and show only roles that weren't in last week's list."

[Claude @claudeai]

Starting to roll out today: Claude Whiteboards, Claude Sheets and Claude Data.

Think a plan through on a board, build a tracker in a sheet, and tell a story with your data.

[Just Samuel @SCrypt22]

@Sergey_Rura Agree. The rule is a filter, not a guarantee. What it buys you is a faster review: with the "say why" line, each new function should come with a reason, so the ones without one are easy to spot.

Do you review every diff by hand, or have an agent do a first pass?

[Just Samuel @SCrypt22]

One rule I'd add to AGENTS.md or CLAUDE.md to cut duplicate code:

"Before writing a new function, search the repo for one that does the job. Reuse or extend it. If nothing fits, say why."

[Just Samuel @SCrypt22]

@aiedge_ love it!

[Just Samuel @SCrypt22]

@simonw The first prompt is the step worth stealing: "Write out a problem.md file describing the problem we want to solve."

That's where Claude found the 2007 BeepBeep paper the whole design rests on, before any code existed.

[Just Samuel @SCrypt22]

@crowdturtle Ask for one section in that report: where they still disagree, with each side's best argument. That's the part I'd read first.

[Just Samuel @SCrypt22]

@witcheer Omg this is cute

[Just Samuel @SCrypt22]

@XFreeze Thanks @elonmusk

[Just Samuel @SCrypt22]

Brian Armstrong runs one lead agent per area, a chief of staff agent over them, and one prioritized queue where every review arrives with screenshots and a video demo.

The queue is the upgrade. Without it, more agents just mean more chats to check.
https://t.co/lTDHHbeI5J

[Brian Armstrong @brian_armstrong]

I've changed how I work over the last few weeks.

I now have one lead agent for each area of focus, running all the time, and a chief of staff agent that oversees them. Each lead manages its own sub-agents that write code, test it, chase reviews, and coordinate with team members.

Everything I need to decide or review comes into one prioritized queue, a simple local web app. Each product review includes screenshots and a video demo of the feature. I clear the queue in batches, mostly by voice and keyboard shortcuts. The leads also bring back ideas I hadn't considered.

Working from one queue, instead of switching between dozens of chats, has cut a lot of mental overhead. The system keeps learning my preferences and distilling my judgment into a "Brian brain," so its recommendations and first drafts get better before they reach me. The goal is fewer corrections per change.

Every change still goes through the same human review and security checks. It is eliminating tasks, not human judgment.

What an exciting time to push the frontier of productivity, especially as we build the economy for agents (agentic finance). We should all keep pushing in this direction.

[Just Samuel @SCrypt22]

@HermesAgentTips @NousResearch One from this morning: Reset Rescue by @brunooliveira. It redeems expiring Codex reset credits across every account in your Hermes pool, with two scheduled tries per credit. It only marks success after the backend confirms.
https://t.co/nqjanewK9g

[Bruno Oliveira @brunooliveira]

I wasn't even planning to enter the Hermes plugin hackathon. Zero ideas.

Then I lost my banked Codex resets. I was squeezing every last drop out of my quota before redeeming them, got distracted, and they expired. Pure rage.

So I built Reset Rescue, a Hermes Agent plugin that redeems your expiring Codex reset credits for you, on every account in your Hermes pool:

- two scheduled tries per credit, 10 and 4 min before expiry
- opt-in early redeem when your weekly quota runs low
- per-account rules and notices on Telegram or any Hermes channel
- success only after the backend confirms it

Built out of hate, so you never lose a reset again.

@NousResearch @Teknium

https://t.co/HfjR2Cj1ZS