Epistemic Stack is an open-source pipeline and web app that ingests sources (including via URL) to build claim-level knowledge graphs for contested questions, with reproducible case-study knowledge bases.
Epistemic Stack is an open-source pipeline and web app that ingests sources (including via URL) to build claim-level knowledge graphs for contested questions, with reproducible case-study knowledge bases.
Project Details
Updated 07/09/26 · By grantmaking.ai · VerifiedI am building Epistemic stack, an opensource tool or rather pipeline for investigating and analyzing contested questions. It perfectly takes massive collections of sources and turns them into claim level knowledge graphs hence making it easier to see where claims come from , how different pieces of evidence relate to one another and which assumptions are driving disagreements. Instead of asking users to trust a conclusion, the goal is to make the underlying evidence and reasoning easier to inspect.
The system is already working. It runs end to end on the COVID-19 origins debate, where it has extracted more than 60 claims linked back to specific source passages and generated reproducible outputs that can be explored through a web application.
I am the only contributor who is working and will be working on the project till completion. I am a senior software engineer with over four years of experience building production software systems.
With funding I will improve the reliability and usability of the existing pipeline, expand it beyond a single case study to many different types of contested questions like COVID-19 origins, a settled science case and an everyday disputed topic, also I will add a URL based ingestion feature so that users can analyze different types of sources without manual preparation, and deploy a public instance that users can use, evaluate and extent.
The project will deliver a web application that is publicly accessible, more than 3 fully developed and publicly browsable knowledge bases, documented and reproducible pipeline runs for each case study, url based source ingestion, and clear documentation so that others can inspect, reproduce and build on top of the work.
Theory of Impact
Updated 07/17/26 · By grantmaking.aiAs AI systems become more capable, there is a growing risk that people will easily accept their conclusions simply because they sound very convincing. A well written answer can feel authoritative regardless of whether is is supported by a strong evidence, weak evidence or no evidence at all. Overtime, this can weaken our ability to independently verify important claims and make informed judgments about what is true. This loss of oversight becomes especially concerning as AI systems play a larger role in research, decision making and public discourse.
Epistemic stack is motivated by the idea that AI should make reasoning more transparent and not opaque. Rather than producing a conclusion and asking users to trust it, the system exposes the evidence and reasoning structure behind that conclusion. Every claim is linked back to the exact source passage it came from, relationship between claims are made explicit and the system analyzes whether support comes from genuinely independent evidence or from the same exact source being repeated through multiple channels.
The main goal is to help people evaluate conclusions on the basis of evidence rather than presentation. Epistemic stack supports this kind of scrutiny and verification that effective human oversight requires by making sources, dependencies and key assumptions visible. In that sense it aims to contribute to a future where increasingly capable AI systems augment our ability to reason about complex questions without asking us to surrender our ability to check their work.
People
Updated 07/17/26 · By grantmaking.aiTeam Member
Discussion
No comments yet. Be the first to share your thoughts.