Ignis

Inspiration

While working on multiple research projects in college, I repeatedly ran into the same frustration: the research workflow is painfully broken. Confirming novelty meant digging through hundreds of papers, experiments dragged on for months, and even after getting results there was always the fear that someone else had already published something similar.

On top of that, writing the paper itself was exhausting, every journal has different formats, obscure requirements, and rigid submission rules. One of my projects took nearly six months end-to-end, which felt absurd in a world where production apps can be built in hours. That frustration became the spark for building Ignis.

What it does

Ignis is an end-to-end AI Research Assistant that compresses the entire research lifecycle into a single workflow.

It:

1. Takes your research idea
   
2. Refines and sharpens it while checking for novelty against existing literature
   
3. Implements the idea, runs experiments, and evaluates results
   
4. Produces a publish-ready research paper automatically
   

The goal is to turn weeks or months of work into hours, without sacrificing rigor.

How we built it

Ignis is built by combining large language models with retrieval systems, experiment automation, and paper-generation pipelines. The system continuously references prior research, translates ideas into executable experiments, and structures results into formal academic writing.

My background working on AI-driven research projects, including AI-driven drug discovery, helped shape the system around real research bottlenecks rather than theoretical ones.

Challenges we ran into

Ensuring novelty checking is reliable and not superficial was a major challenge, research demands precision. Automating experiments while keeping them reproducible and scientifically sound was another challenge. Finally, generating papers that meet strict academic standards (instead of generic summaries) required careful design and iteration.

Accomplishments that we're proud of

We validated demand early by building a waitlist of 120+ users, many of them active researchers and students. Nearly everyone we spoke to echoed the same pain points: slow literature reviews, long experiment cycles, and frustrating paper writing. That consistent feedback confirmed we’re solving a real and deeply felt problem.

What we learned

Researchers don’t just want faster tools, they want tools they can trust. Accuracy, transparency, and reproducibility matter far more than flashy features. We also learned that research workflows are surprisingly similar across domains, making end-to-end automation far more powerful than we initially expected.

What's next for Ignis

Next, we’re focused on improving experiment reliability, deeper novelty analysis, and supporting more research domains. Our long-term vision is to make high-quality research dramatically more accessible,

Built With

• langchain
• mastra
• node.js
• python
• react