Analytics platform powered by LLM and decentralized fact-checking | MOBR Systems F15 Proposals

[mobr.ai]

Hello Cardano community,

We would like to share two proposals that MOBR Systems has submitted to Project Catalyst’s Fund 15, both focused on strengthening Cardano’s data, governance, and trust infrastructure.

Our proposals received solid evaluations from reviewers. CAP (Cardano Analytics Platform) with an overall score of 4.27, while d-FCT (Decentralized Fact Checking Tool) received 4.49, placing both in a strong position for this fund.

Cardano Analytics Platform powered by LLM - MVP

The problem CAP addresses is fundamental. Cardano produces a large amount of on-chain and off-chain data, yet accessing, relating, understanding, and trusting that data is quite difficult for many users, including technical and non-technical. Information is often fragmented across explorers, dashboards, and raw APIs, with little semantic structure.

CAP solves this by providing an analytics platform built on a Cardano knowledge graph, combining structured blockchain data with natural language querying and smart dashboards. This enables clearer and easier insights into network activity and ecosystem trends. Better analytics directly supports more informed decisions, higher transparency, and broader participation across the ecosystem.

F15 proposal here

d-FCT: Decentralized Fact-Checking Toolkit (Phase 2)

The problem d-FCT addresses is the lack of reliable and transparent fact-checking, where current processes are slow, inconsistent, often biased, and difficult to scale. As AI generated content becomes more common, distinguishing verified claims from misinformation becomes harder, even for experienced eyes.

d-FCT introduces a decentralized fact-checking framework built on Cardano, where claims, evidence, and verification steps are transparently recorded and traceable. It combines on-chain incentives and AI-assisted analysis to support community-driven verification without relying on a single authority. It is a new category of high-impact, public-good infrastructure built on Cardano and designed to drive real-world engagement, on-chain activity, and ecosystem expansion. The return on investment extends across technical, social, economic, and ecosystem-development dimensions.

F15 proposal here

Why this matters?

Both proposals aim at delivering practical infrastructure that can be reused by the wider Cardano ecosystem.

If you believe Cardano benefits from better analytics, stronger tooling, and more transparent information flows, we invite you to take a look at the proposals and consider supporting them in Fund 15.

Thank you for your time and for participating in Catalyst.

[mwatsimulamo]

I have just read these proposals carefully, and based on my understanding, the first one—which aims to provide a Cardano knowledge-graph–based analytics platform, combining structured on-chain data with natural-language queries and intelligent dashboards—seems quite clear and relevant. It effectively adds value to the Cardano ecosystem by offering the community more accessible tools to easily explore on-chain information, complementing what existing explorers and similar tools already provide.

However, I do have a concern regarding the second proposal. Could you please elaborate a bit more on the problem you are addressing, specifically the issue of the lack of reliable and transparent fact-verification? In your post, you mention that current processes are slow, inconsistent, often biased, and difficult to generalize. I would appreciate a clearer explanation of this challenge, as I personally find it difficult to fully grasp the problem in its current form.

That said, both proposals are well written, and I would genuinely be happy to support you during the voting process.

[mobr.ai]

Hi Olivier. Thank you very much for taking the time to read both proposals so carefully and for the thoughtful feedback. We really appreciate it, and we are glad the value of the analytics platform came across clearly.

Regarding the second proposal, d-FCT, the core problem we are addressing is not the absence of fact-checking per se, but how fact-checking currently works in practice and why it fails to scale or earn broad trust.

Today, most fact-checking processes are centralized, opaque, and manual. Decisions are often published as final verdicts without a clear, inspectable trail showing how conclusions were reached, which sources were used, what evidence supports or contradicts a claim, and where uncertainty still exists. This makes results hard to audit, hard to reproduce, and difficult to reuse across contexts. Two different organizations may check the same claim and reach different conclusions, with no structured way to compare reasoning, evidence quality, or assumptions.

In addition, contributors to fact-checking efforts are usually volunteers or part of closed teams. There are few transparent incentives for broader participation, and no reliable way to attribute, reward, or evaluate individual contributions over time. This limits both scale and diversity of perspectives, especially for fast-moving or niche topics.

d-FCT proposes a different approach. Instead of producing a single “verdict”, it aims at structuring the entire verification process as an open, inspectable graph of claims, evidence, sources, and reasoning steps. This way, each claim could be traced back to concrete evidence artifacts, each piece of evidence has provenance, and disagreements or contradictions are explicitly represented rather than hidden. This makes verification outcomes explainable and comparable, rather than authoritative black boxes.

Importantly, the goal is not to replace human judgment with automation, but to make verification more transparent, strutured, and reusable, while using LLMs and analytics to reduce friction in tasks like claim extraction, evidence discovery, and summarization. Over time, this enables decentralized participation with clearer accountability and measurable quality signals.

We hope this clarifies the problem space and why we believe it is still largely unsolved at an ecosystem level. Thank you again for the encouragement and for your willingness to support us during the voting phase. It truly means a lot.

[rbrandao]

Thanks for your support, @mwatsimulamo.

The video shared here walks through the project vision and may help provide additional context: https://x.com/mobrsys/status/2011455693753291096

[mwatsimulamo]

I now have some understanding of the issue, and I believe we can continue this discussion further—perhaps even via Telegram—for more detailed explanations, especially since fact-checking is involved.
In my own point of view and understanding, while AI or system automation can be very useful, it can also have limitations in reasoning and context, which sometimes makes the involvement of third parties—such as administrators or moderators—necessary to provide informed judgment beyond automation alone.

Indeed, stay encouraged, and I am genuinely pleased that your proposals aim to address problems that go beyond surface-level considerations and simple reflections.

[mwatsimulamo]

This video certainly sheds more light on the project, and it is indeed a very interesting initiative. That said, it may be worth updating it, as the video currently outlines milestones for the year 2025, whereas we are now in 2026.

Wishing you continued strength and success with this excellent project.

[Marcio_Moreno]

Hi @mwatsimulamo. Thanks for your interest. The 2025 milestones were the ones we accomplished, while the others are the ones we specified to achieve in 2026.