Founder thesis · Spring 2026
Shared Intelligence is the next collaboration layer.
Private AI made individuals faster. The harder, more interesting move is shared context - where people and agents reason about the same work, in the same room, with visible memory, real human authority, and bounded follow-through.
Start with the manifesto
I build companies that change how people work.
AEFIS spent a decade rewiring how universities assess student learning. We bootstrapped it from three founders into a market-leading EdTech platform serving 55 institutions, and exited to private equity in 2021.
Sociail is the second act - a shared AI workspace where humans and agents stay in the same room, drafting against the same memory, deciding against the same history. This site is where I think out loud about the category we're building toward: Shared Intelligence.
Selected results. AEFIS bootstrapped 2012, scaled to 55+ universities, 4.8x return at PE exit in 2021. Philadelphia 100 in 2016 and 2017. Drexel commerce and engineering. Bilingual EN / TR.
Sociail - a shared room for humans and agents.
Drafts, decisions, memory, and approvals live in one place, not split across a dozen tabs and a hundred prompts. Agents appear as first-class participants in the room - visible, named, accountable - not as a tool you reach for and then forget you used.
The wager is simple: most of the value of AI at work isn't faster solo prompting. It's whether your team's context, judgment, and follow-through compound across the people and agents doing the work. We're building for that.
Recent thinking
Notes on AI-native work, trust, and the operator path behind Sociail.
- 06 May / 2026
Shared Intelligence Starts With a Shared Room
Private AI made individuals faster, but work still fragments across tabs, meetings, and memory. Shared Intelligence starts with a simple shift: people and AI agents working from the same visible, permissioned room.
- 02 May / 2026
The Productivity Singularity Is a Workflow Problem
AI is collapsing the distance between intent and execution. The real question is whether responsibility is designed into the workflow before action happens.
- 01 May / 2026
What Trustworthy AI Collaboration Looks Like
Trustworthy AI collaboration is not about trusting the model. It is about designing the system around context, roles, human authority, memory, and accountability.
Team
My team is my superpower.
Pictured here with the AEFIS team just days before the COVID-19 lockdown. A difficult, transformative time - but also one of our most inspiring chapters. I am forever grateful to those who helped shape that journey.

We don't just think, therefore we are. We share intelligence, therefore we become.Mustafa Sualp