Skip to content
View abhichat85's full-sized avatar
🎯
Building AI + SaaS + Agency
🎯
Building AI + SaaS + Agency
25 followers · 15 following

Achievements

Achievement: YOLOAchievement: Pair ExtraordinaireAchievement: Pull SharkAchievement: Quickdraw

Achievements

Achievement: YOLOAchievement: Pair ExtraordinaireAchievement: Pull SharkAchievement: Quickdraw

Block or report abhichat85

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
abhichat85/README.md

⚡ What I believe

AI agents will replace everyone in the knowledge economy. Not a prediction — already happening. Most people are still sleeping on it.

I build the infrastructure for the companies making that transition real.

⭐ Highlights

2026 — Flagship OSS

agent-stream
agent-stream
The SSE event protocol for AI agents. Extracted from 36 tools, thousands of production runs.

pip install agent-event-stream  ·  npm install @agent-stream/react 		2026 — New · Rust

ANT
ANT — Agent-Native Terminal
Like Warp, redesigned from first principles for AI agents. Approval-gated. Never runs arbitrary shell commands.
Why SSE for AI agents keeps breaking at 2am — DEV.to The four bugs every team hits. The protocol that fixes them.
I've burnt 43 Million tokens in Claude Code — Medium What building at this scale actually teaches you.
The Anatomy of an Agent Harness — Medium On the engineering that lives between a rented model and a user who trusts it.

🏗️ Einstein Labs — Operating Stack for AI-Native Companies

Six products. One thesis: AI-native companies need infrastructure built for them — not San Francisco pricing and assumptions.

Praxiom AI
Praxiom AI
Autonomous AI product manager. 36 agent tools shipped. Multi-agent harness, overnight research cycles, quality-scored PRDs. Integrates with Linear, GitHub, Jira. 		Polaris
Polaris
Spec-driven AI coding agent. Real agent in E2B sandbox. Sees its own runtime errors. Raw Anthropic SDK — no abstraction wrappers. Built for India + SEA.
Pulse AI
Pulse AI
Product intelligence engine. LLM-powered behavioral analytics. The layer between your event stream and a decision a PM can act on. 		Prism
Prism
Autonomous content intelligence. Multi-tenant AI-native content OS. Generates at scale, maintains brand voice across channels. Runs independently.
Pylon
Pylon
Billing infrastructure for AI companies. Per-token, per-agent-run, per-insight pricing. Not Stripe bolted on top — built for the AI billing model from scratch. 		Parallax
Parallax
Autonomous product video OS. Give it your product URL, screen recordings, and a goal — it returns a complete multi-variant video campaign. Cursor for product videos.

🔧 Open Source

Tools built from production, not from ideas.

agent-stream
agent-stream
The SSE event protocol for AI agents 

token · thinking · tool_use · tool_result
turn · progress · creation · done · error

Python emitter (FastAPI/Starlette)
React hook: useAgentStream
AgentStreamRecorder: JSONL + CLI replay
Zero external dependencies

pip install agent-event-stream
npm install @agent-stream/react

 ANT
ANT — Agent-Native Terminal
AI-aware CLI, built in Rust 

ant diagnose tests
→ detects test runner
→ captures output
→ LLM explains failures

Shows: command · cwd · risk level
before running anything.
Run history: ~/.ant/ant.db
pmeval Benchmark for AI PM agents. 336 tasks, 14 categories, LLM judge, hallucination detection.
knights-planning-core Manus-style persistent markdown planning for Claude Code.

📝 Writing

Production notes from building AI infrastructure at scale.
The Anatomy of an Agent Harness May 2026 · On the engineering that lives between a rented model and a user who trusts it
I've burnt 43 Million tokens in Claude Code Apr 2026 · What building at this scale actually teaches you
Why SSE for AI agents keeps breaking at 2am Mar 2026 · The four bugs every team hits — and the protocol that fixes them
The Last Human Advantage: Judgement, Taste, and the 10x Question Mar 2026 · What remains uniquely human when AI handles everything else
Starting Up in the Age of AI Jun 2025 · The startup game has changed. What that means for founders.

🛠️ Stack

Philosophy: Raw SDK over abstraction wrappers. If you don't know what's underneath, you will at 2am.

📊 By the numbers
🛠️ 36 Agent tools shipped in production
🔥 43M Tokens burned in Claude Code
📋 336 PM benchmark tasks (PMEval)
🏗️ 6 Products in the Einstein Labs P-family
✍️ 17+ Years shipping software

LinkedIn · 12K+  ·  Medium  ·  praxiomai.xyz  ·  X / Twitter  ·  Email

Pinned Loading

  1. agent-stream agent-stream Public

    The SSE event protocol for AI agents — token, tool_use, thinking, progress, done. Python emitter + React hook.

    Python 1

  2. abhishek-portfolio-website abhishek-portfolio-website Public

    Interactive Terminal OS portfolio — dark terminal aesthetic, IDE layout, sidebar navigation. Built with Next.js 15, TypeScript, Tailwind CSS, Framer Motion.

    TypeScript

  3. abhichat85 abhichat85 Public

    GitHub profile README

  4. ant ant Public

    ANT: Agent-Native Terminal — run AI agent definitions from the command line

    Rust