Founders, Inc. & LocalHost HQ Case Study
Founders, Inc. & LocalHost HQ Case Study
Founders, Inc. & LocalHost HQ
Founders, Inc. & LocalHost HQ
Founders, Inc. and the LocalHost India Founders Program selected me to build Macy as a Founder in Residence. Everything under that brief was mine to build: the always-on agent layer watching inbox, calendar and markets, the safety model for autonomous actions, then the product and go-to-market on top.
Founders, Inc. and the LocalHost India Founders Program selected me to build Macy as a Founder in Residence. Everything under that brief was mine to build: the always-on agent layer watching inbox, calendar and markets, the safety model for autonomous actions, then the product and go-to-market on top.
2025
Founder in Residence
Founder in Residence, 2025 · Founders, Inc. and the LocalHost India Founders Program · Bengaluru, India
The product built during the program is Macy, a proactive personal AI agent that runs inside iMessage. It is live at heymacy.ai.
Mission
Mission
I joined the program to answer one question: can an AI agent be genuinely proactive without being reckless? A useful agent has to act before it’s asked, watching your inbox and calendar on its own initiative, and still earn the right to touch anything that matters.
I joined the program to answer one question: can an AI agent be genuinely proactive without being reckless? A useful agent has to act before it’s asked, watching your inbox and calendar on its own initiative, and still earn the right to touch anything that matters.
Challenge
Challenge
A proactive agent has to be awake when the user isn’t. Running background jobs against inbox, calendar and markets 24/7, across Gmail, Calendar, Drive, Slack, WhatsApp and 500+ apps, meant a persistent layer that decides unprompted whether something is worth a text, and never acts on judgment alone.
A proactive agent has to be awake when the user isn’t. Running background jobs against inbox, calendar and markets 24/7, across Gmail, Calendar, Drive, Slack, WhatsApp and 500+ apps, meant a persistent layer that decides unprompted whether something is worth a text, and never acts on judgment alone.
Solution
Solution
I architected an always-on agent layer with background jobs watching those sources around the clock, then paired it with an approval-gated action model: nothing sensitive fires without an explicit reply, and each action lands in a full audit trail. I led product and go-to-market on top of that architecture.
I architected an always-on agent layer with background jobs watching those sources around the clock, then paired it with an approval-gated action model: nothing sensitive fires without an explicit reply, and each action lands in a full audit trail. I led product and go-to-market on top of that architecture.
Result
Result
Macy came out of the program as a working proactive agent rather than a prototype, with a 150+ pre-launch waitlist behind it. I owned that outcome end to end as Founder in Residence: the architecture, the safety model, the product and the go-to-market.
Macy came out of the program as a working proactive agent rather than a prototype, with a 150+ pre-launch waitlist behind it. I owned that outcome end to end as Founder in Residence: the architecture, the safety model, the product and the go-to-market.