Thesis
Agents are the next users of software.
We're building their world.
The web was built for humans. Every interface, auth flow, and feedback loop assumes a person at the keyboard. Agents need different primitives. We build them — and we build the agents that prove they work.
Read our vision →Doctrines
08UI Parity
Whatever the user can do, the agent should do via tools.
Atomic Tools
Tools are primitives. Features are agent loops.
Prompts as Features
New capabilities via new prompts, not code.
Emergent Capability
The agent does things you didn't design for.
Compounding Context
Apps improve through context and prompts, not just code.
Single Agent > Swarm
One well-equipped agent beats a fleet of fragile ones.
Build-Learn-Build
Build products. Discover gaps. Build that layer. Repeat.
The Web Isn't Agent-Ready
Current software assumes a human operator. That's the gap.
Projects
07niahere
A personal AI agent you fork and make your own. Small enough to fully understand, no abstractions to fight. Extend through skills that transform code, not config. The opposite of a black box.
Multi-agent team orchestration. 2nd place at Codex PeakXV Hackathon (Feb 2026) on the multi-agent track.
Calculate LLM API token costs. Python + JavaScript library and CLI. Deterministic pricing with local cache.
CLI that reads local AI agent session files, calculates API costs, and generates a unified dashboard. Supports Claude Code, Codex, OpenCode, Amp.
Cloud dashboard for code-usage. Multi-device sync, leaderboard, trend tracking. Pair any machine and see your AI spend from anywhere.
Curated library of optimized prompts for ChatGPT, Claude, Gemini, and other AI tools. Browse 20+ categories or request custom prompts for specific tasks.
AI-native stock research platform for Indian equities. Source-backed NSE/BSE market data, fundamentals, corporate actions, filings, index exposure, sector breadth, and technical signals in one research surface.
Impact
All numbers →$0.00T
AI spending this year
$0B
Anthropic ARR, Apr 2026
0M
MCP downloads/month