Data analyst who builds AI agents that run without me. I post what it cost and what broke. Repos are public.
California, USA
Joined January 2026
- Tweets2K
- Following749
- Followers235
- Likes789
Nitesh retweeted
New in Claude for Legal: a gallery of ready-made artifacts for contract review, regulatory horizon scans, cite checks, and more.
Start from the built-in example, add your own documents and connectors, and share it with anyone across the business.
Sometimes it feels like you're overcomplicating an AI product when a simple comparison site could help people choose the right automation tool and earn from every referral.
Your coding agent dies when you close the laptop.
Orca keeps working on the train.
It runs a fleet of coding agents in parallel,
on your own model subscription.
Desktop, phone, or a VPS that never sleeps.
Two things make it worth a look.
It is your subscription, not a new bill.
Point it at the Claude or Codex plan you already pay for.
No per-seat markup between you and the model.
And it runs many at once.
Kick off five agents on five branches.
Check them from your phone at lunch.
Merge the ones that landed.
The work does not wait for you to be sitting in front of it.
That is the shift underneath it.
Watching one agent type is not the job anymore.
Starting several and reviewing what they finished is.
MIT licensed and open source.
53,000 stars.
Your AI content used to vanish when the feed moved on. But search can give useful posts a longer life.
When you build content around real questions, you can track which topics appear in search, which terms bring clicks, and where your audience is finding you.
And that changes how you think about posting consistently.
Because a quick AI tip may get attention for a few hours. Then it gets buried.
But a clear answer to a specific problem can keep attracting the right people for much longer.
And this matters even more when your audience is searching for tools, workflows, and practical comparisons.
The challenge is that most AI content is written for the feed, not for the question behind the search.
So start with a problem people already have: choosing an automation tool, running a model locally, or checking whether an AI answer is reliable.
Use search suggestions, community discussions, and question tools to find the exact language people use.
Then answer one focused question in every post you publish.
Because each post can become more than content. It can become a useful discovery point for your audience and a natural place to recommend a relevant tool when it genuinely helps.
AI companies are moving toward clearer labels for machine-generated content.
Just wait—the platforms will adapt around those signals too.
And when they do, marketers publishing raw AI output may find distribution harder.
Generic AI content already blends into the feed… labels will make that pattern even easier to spot.
If you’re building an AI content workflow, don’t stop at generation.
Add real examples, useful analysis, and human judgment before publishing.
One resume
Multiple billion-dollar companies.
Rockets, Cars, AI,Space, Payments and more
Which one is the most impressive? 👇
The best automation tool is not the one with the most integrations.
It is the one that lets you see exactly where the workflow made a decision.
Hot take:
$300 for Grok Heavy. $200 for ChatGPT Pro. $200 for Cursor Ultra.
$8,400 a year. For one person. On plans they call consumer.
That is enterprise pricing in a hoodie.
Everyone posts RAG, Agents, MCP, and A2A like a menu you pick from. They are not a menu. They are a stack.
RAG gives a model knowledge it was never trained on.
An agent wraps that model in a loop so it can act, not just answer.
MCP is the standard socket that agent uses to reach its tools.
A2A is how one agent hands work to another.
Four different jobs, one system. Read them as competitors and you burn weeks asking which one to use, when a real agent usually needs all four at once, each doing the job the others cannot. Read them as layers and you always know where each one belongs.
Made the diagram so the whole thing fits in one look.
You do not need 100 tutorials to understand AI agents. You need 8 videos, watched in order, from the people who actually build them.
The first 3 explain the model. The last 5 build the agent. Skip the order and nothing sticks.
Saved you the search. Every link below 👇
7. Let's Build an Agent from Scratch, Kam Lasater.
No framework, just the loop. Watch it once and you stop believing there is magic inside the abstractions.
youtube.com/watch?v=xzXdLRUy…
8. PhiloAgents, a full build playlist by The Neural Maze.
Multi part and hands on, where you put all of it together into something real.
Start at the top, one a day, and you understand agents by next week.
youtube.com/playlist?list=PL…
