Synic

Inspiration

As electronics engineering students, we were trapped between managing complex capstone research and mapping out early-stage startup ideas, completely drowning in unorganized journal PDFs, sensor data matrices, and chaotic brainstorms. Traditional AI tools were far too agreeable; they simply formatted messy notes and praised ideas instead of exposing structural flaws. This inspired Synic—a strategic devil's advocate built to aggressively stress-test raw thinking before a market or a thesis committee does.

What it does

Synic is an AI-powered workspace that transforms unstructured data into a validated action plan. Users dump raw notes into "The Chaos Bin" on the left, while "The Sparring Ring" on the right generates an objective critique. The system maps intent against evidence, drops high-visibility alert cards on logical flaws, and outputs a pressure-tested action plan complete with a sharpened core thesis and immediate validation tasks.

How we built it

The application is built as a responsive, dark-mode single-page web interface using HTML, Tailwind CSS, and Lucide Icons. The intelligence layer uses a custom system prompt matrix optimized for low-temperature inference inside Google AI Studio, forcing the model to bypass polite filler text. Vanilla JavaScript drives the real-time interactive toggling between "Founder Mode" and "Student Researcher Mode" to alter evaluation parameters instantly.

Challenges we ran into

The primary hurdle was engineering the "Friction Alignment." Standard conversational prompts made the AI too passive, while over-cranking the critique made it destructively pedantic. The solution required strict operational directives built into the system instructions, explicitly forbidding generic praise and mandating that every piece of critical pushback tie back directly to an empirical data gap or a system constraint.

Accomplishments that we're proud of

We successfully shifted AI from a passive notebook into an active intellectual counterweight that refuses to be a "yes-man." We built a zero-install prototype that delivers biting, highly contextual critiques within seconds, seamlessly pivoting from catching market validation biases for a founder to flagging methodology contradictions for a researche

What we learned

Building Synic taught me to view project management through objective risk assessment rather than emotional optimism. If we treat an idea's validation as a state vector of unvalidated assumptions x(t), we can model risk reduction as a closed-loop control system. When a project relies on assumption over empirical data, the risk factor R grows exponentially:

$$R = \int_{0}^{T} \left( \frac{A(t)}{F(t) + \epsilon} \right) dt$$

Where A(t) represents speculative assertions, F(t) represents validated baseline facts, and \epsilon is a small stabilizing constant. Synic minimizes R by forcing the user to expand their empirical foundation.

What's next for Synic

The next phase is integrating live vector embeddings (RAG) so students can upload entire folders of research PDFs for real-time cross-examination. We also plan to build a browser extension to offer an instant "Spar" button while drafting text online, and an automated tool that converts identified blind spots directly into customer discovery outreach templates.