Inspiration
AI agents are getting better at making decisions, but in regulated industries, being “smart” isn’t enough. We need to know
why an agent acted, what it was allowed to do, and whether it actually followed the rules.
That led us to Aegis: a trust layer for agentic systems in financial services.
What it does
Aegis lets AI agents reason and collaborate while keeping them inside clearly defined boundaries.
It checks actions against policies before execution, records what happened, and creates a verifiable trail of who decided what, why, and what was actually executed.
The goal is simple: make autonomous AI feel controlled, explainable, and trustworthy.
How we built it
We built Aegis around three layers: Reason, Guard, Prove.
Agents handle the reasoning and workflow. A policy layer determines what they are allowed to do. A verification layer captures the decision and execution trail so every important action can be audited later.
Instead of replacing existing systems, Aegis sits between agents and the systems they interact with.
Challenges we ran into
The hardest part wasn't making agents take action. It was deciding how much freedom to give them.
Too much control makes agents useless. Too little creates unacceptable risk.
We spent a lot of time thinking about policy boundaries, multi-agent decisions, exceptions, and how to prove that an action actually followed the rules.
Accomplishments that we're proud of
We moved beyond the idea of “AI that automates compliance” and built around a different question:
Can we make autonomous AI trustworthy enough for regulated environments?
Aegis gives every important decision a boundary, a reason, and an evidence trail.
That's the foundation we wanted to prove.
What we learned
Agentic AI isn't just about making agents more autonomous.
The real challenge is building the trust infrastructure around that autonomy.
We learned that policies shouldn't simply stop agents. They should give agents a safe operating boundary so they can move faster without losing control.
What's next for Aegis: Trust, protection, institutional
Aegis starts with trust.
Next, we want to build deeper protection — stronger policy enforcement, continuous monitoring, risk detection, and intervention when agents behave outside their boundaries.
Eventually, we see Aegis becoming an institutional trust layer for agentic systems: something financial institutions can use to safely deploy autonomous agents across onboarding, compliance, wealth management, operations, and beyond.
Agents can provide the intelligence. Aegis provides the guardrails and proof.
Log in or sign up for Devpost to join the conversation.