Jeremiah Agthe

ai agents || need a custom agent? → your work that you can focus to make more money 💰

[Jeremiah Agthe @Serial_Builder]

I’ve been talking with a nine-figure e-commerce business owner on leveraging AI and automation.

Here are some key insights on where e-commerce businesses can score quick wins:

1. They’re investing $122k/month in content creation and prioritized assistance with researching TikTok trends and streamlining static ad variations.

2. We uncovered 287 comments on a competitor’s TikTok post detailing why customers switched brands —> far more valuable than any traditional survey.

3. Their creative team spends hours manually browsing TikTok, Instagram, and Reddit for ideas. We’re automating this process to cut it down to minutes.

4. AI is being used to generate 90% of the initial work for static image iterations, allowing their team to refine rather than build from the ground up, saving significant time.

5. They’re not deploying AI to cut jobs but to free up their talent from repetitive, time-consuming tasks.

The biggest gains aren’t from complex n8n automations but from straightforward systems that save their teams hours of effort every week.

[Jeremiah Agthe @Serial_Builder]

life is about retardmaxxing and between being rational

[Jeremiah Agthe @Serial_Builder]

@iatnon @ClaudeDevs nah, its u

[Jeremiah Agthe @Serial_Builder]

@NoahRyanCo Breathwork/meditation, enough sleep, working out, eating real food. Pretty basics

[Jeremiah Agthe @Serial_Builder]

@iatnon @ClaudeDevs I dont have any issues 👀

[Jeremiah Agthe @Serial_Builder]

@Sushilk91 openAi is pushing narartive here... it so redicilous to see people blasting claude code, opus 4.8 even tho its still the best tool out there -> or then everyone has just a huge skill issues xd

[Jeremiah Agthe @Serial_Builder]

please spawn as many subagents u want and burn as mouch tokens as u want if it makes the job.

[ClaudeDevs @ClaudeDevs]

We've reset 5-hour and weekly rate limits for all users on Pro and Max plans.

We fixed an issue that caused some Claude Code sessions to spawn excessive parallel subagents, burning through usage faster than expected.

[Jeremiah Agthe @Serial_Builder]

@mtmalinen which paid AI u used? i think u have a skill issue ps. use claude opus 4.8 and the max 20x plan (270€/kk) with claude code and u will change ur opinion 100%

[Jeremiah Agthe @Serial_Builder]

@giordanorandone And repeat till we die.

[Jeremiah Agthe @Serial_Builder]

This.

[Alex Becker 🍊🏆🥇 @ZssBecker]

Drinking ALWAYS come at a cost.

Optimize your health. Go sober for 30 days. Even "just two" dramatically drops your baseline mood/gusto for days.

Does it mean give up drinking? No. It has social and creative benefits.

Just stop lying to yourself about the cost. It's poison.

[Jeremiah Agthe @Serial_Builder]

Same in ASIA/SEA…

In Finland that is almost the first thing people ask lol

[Carnivore Aurelius ©🥩 ☀️🦙 @AlpacaAurelius]

The best part about living in Latin America was that people didn’t ask “what do you do” within 30 seconds of meeting

[Jeremiah Agthe @Serial_Builder]

@dangreenheck Yeah its just u

[Jeremiah Agthe @Serial_Builder]

i anyone actually using any other model than opus 4.8 with ultracode? why would u even use the low effort models?

[Jeremiah Agthe @Serial_Builder]

@hqmank lol not even close

[Jeremiah Agthe @Serial_Builder]

@jakobgreenfeld i smell skill issues

[Jeremiah Agthe @Serial_Builder]

i'm a top 5% claude code user in the world.

Here is how I build multiple projects at the same time using Claude Code without burning out cognitively...

1. Give a clear end goal for each terminal/project.
2. Pause each session at 150k-250k range.
3. Use a /handoff skill.
4. /clear the session.
5. Paste the handoff prompt on fresh session.

repeat as long as the end goal is reached….

Use Opus 4.8 + ultracode + ultrathink + auto-mode

Paste this prompt for the /handoff (+ /tokens skill):

"Build me two Claude Code skills, then tell me to restart it:

 1. ~/.claude/skills/tokens/SKILL.md — command /tokens. Reads my live session
 transcript (newest *.jsonl in ~/.claude/projects/<cwd-slug>/), sums the last
 assistant message's input + cache_read + cache_creation tokens, and reports
 context used vs window (200k, or 1M for 1M models) as a % + bar — warns past 75% (wrap up) and 90% (run /handoff now).

 2. ~/.claude/skills/handoff/SKILL.md — command /handoff. FIRST verifies real
 state (git status, git log, git diff --stat, build/test) instead of trusting
 the chat. Then writes a fresh-session continuation prompt: goal, verified
 state, what's done, what's left, key files as path:line, gotchas, exact next
 step, run/verify commands. Saves to ~/.claude/handoff/<project>.md and prints it."

[Jeremiah Agthe @Serial_Builder]

Opus 4.8 is BETTER than GPT-5.5 at EVERYTHING!

There's no need to lie

[Angel 🌼 @Angaisb_]

Opus 4.8 is NOT better than GPT-5.5 at coding

There's no need to lie

[Jeremiah Agthe @Serial_Builder]

"Everyone has AI now."

thats total bs. we live in a bubble.

[Umair Shaikh @1Umairshaikh]

Everyone has AI now.

That advantage expired fast.

What's new advantage?

[Jeremiah Agthe @Serial_Builder]

@sflorimm technically yes -&gt; claude does

it ran 149 agents with ultracode + opus 4.8 for one hour

[Jeremiah Agthe @Serial_Builder]

Note: u dont need to use word workflow. Claude is smart enough to launch the dynamic workflows if the task ”require” it…

[Min Choi @minchoi]

Claude Code has a new wild feature... dynamic workflow.

How to try:

set /model to opus 4.8
set /effort to "ultracode" 😂
use “workflow” in your prompt

Claude will write an orchestration script, spawn subagents swarm, verify results, and report back.