
Nihal Rahman
21 y/o (unc)
i can be whatever you : ai engineer, devops guy, full stack dev :) or an overly enthusiastic tokenmaxxing fun guy to have at the office :)
currently making law fun
Fed10 (YC W26)
prev founding eng
ghost (a16z)- jarvis. For REAL.
co-founded
rae - cool desktop buddy/agent
started an nihal.iscooked around my startup journey, peaked at 1.5m+ monthly views
i love giving back to the world: furnace,furnace, perch,perch, flowlensflowlens
i like building things that feel alive. obsessed with making ai actually useful, not just cool demos. when i'm not coding, i'm probably breaking something trying to make it better.
either that or prolly busy obsessing about something i found cool last night.
also grinding skateboarding/football·coding @ treks/beaches
making legal work faster and clearer. i'm shaping the agent experience, product systems, and the small interactions that make it click.
working across product, interaction, tool use, approvals, context, and orchestration.
learning how to deal with a **** ton of data while keeping the model grounded, fast, and actually useful.
spent a year learning how to clone “you”, understanding your behavior across every app, how you act, and how you text.
built local-first memory systems and harness optimizations that could carry that context across your computer.
pushed the frontier of browser and computer use so ghost could act across your apps and do it all for you.
a desktop-first, on-screen buddy built just as ai started going mainstream.
one app for all your models, tools, and memory, wrapped into something approachable and genuinely easy to use.
grew to a 1,000+ person waitlist before the team was acqui-hired by ghost.
a from-scratch agent harness built from the best parts of modern coding agents, without the baggage.
graph-based session forks, typed tools, permission gates, skills, parallel subagents, plan mode, and provider switching.
headroom compresses huge tool outputs without losing them; /evolve can modify, verify, and carry local harness changes across updates.
a macOS companion that turns the notch into a home for ai, apps, tasks, alerts, and system stats.
one tiny surface that stays out of the way until you actually need it.
an experiment in ambient assistance without the distraction.
a desktop debugger that puts frontend and backend behavior into one continuous trace with almost no setup.
it groups the dom, network, console, react state, backend spans, and source maps.
paste a url, reproduce the bug, and stop chasing scattered logs.
i like building things that feel alive. obsessed with making ai actually useful, not just cool demos. when i'm not coding, i'm probably breaking something trying to make it better.
either that or prolly busy obsessing about something i found cool last night.
also grinding skateboarding/football·coding @ treks/beaches
making legal work faster and clearer. i'm shaping the agent experience, product systems, and the small interactions that make it click.
working across product, interaction, tool use, approvals, context, and orchestration.
learning how to deal with a **** ton of data while keeping the model grounded, fast, and actually useful.
spent a year learning how to clone “you”, understanding your behavior across every app, how you act, and how you text.
built local-first memory systems and harness optimizations that could carry that context across your computer.
pushed the frontier of browser and computer use so ghost could act across your apps and do it all for you.
a desktop-first, on-screen buddy built just as ai started going mainstream.
one app for all your models, tools, and memory, wrapped into something approachable and genuinely easy to use.
grew to a 1,000+ person waitlist before the team was acqui-hired by ghost.
a from-scratch agent harness built from the best parts of modern coding agents, without the baggage.
graph-based session forks, typed tools, permission gates, skills, parallel subagents, plan mode, and provider switching.
headroom compresses huge tool outputs without losing them; /evolve can modify, verify, and carry local harness changes across updates.
a macOS companion that turns the notch into a home for ai, apps, tasks, alerts, and system stats.
one tiny surface that stays out of the way until you actually need it.
an experiment in ambient assistance without the distraction.
a desktop debugger that puts frontend and backend behavior into one continuous trace with almost no setup.
it groups the dom, network, console, react state, backend spans, and source maps.
paste a url, reproduce the bug, and stop chasing scattered logs.



p.s. my favourite kind of farming is token farming