Hey everyone,
Happy the share this Data Flow across Trust Boundaries using Decentralised Ledgers article.
Explores the use of Cardano and the Cardano Foundation’s Reeve - Ledger on Blockchain - Platform (https://github.com/cardano-foundation/cf-reeve-platform).
There’s also a Spring / Kotlin app to go with it, that implements and extends the Reeve platform as well as the follower app.
https://github.com/Edgxtech/cms-demo-app
Might find it interesting or useful
If you are the author, can you please post the actual content so that it can be read directly without needing an account on another site?
Ok, this is the link can use if not a Medium member or don’t have an account
It’s also in the intro
”Read on here if you can’t view below.”
Hey Tim, fantastic article!
The Cardano community is in dire need of this kind of in-depth, practical content. I was inspired by your consignment tracking demo, which is a fantastic illustration of what is feasible. I also looked at the GitHub repository, which led to some further detailed follow-up inquiries:
Regarding the Data Challenge of “Pre-Reeve”: A crucial point is brought to light by your own CMS: data must be properly formatted before Reeve can process it. Do you think the difficulty of organizing their “messy” real-world company data into a clear, Reeve-compatible format is a significant obstacle to adoption for the many firms that lack a current ERP or a developer to create a custom app?
On the Friction of the Toolset: What, in your opinion, are the most important procedures or abstractions required to enable a small-to-medium business without a professional DevOps staff to use the toolset itself (a local node + Reeve)?
In reference to the Indexer Extension: The extension of the follower-app to index consignment-specific events was the main change, according to the README. To what extent is this YaciStore based indexer expandable from an architecture perspective? Would each new use case necessitate a substantial amount of unique development work, or could it be readily modified to listen for and index various kinds of on-chain events, such as multi-sig treasury transactions or DAO governance votes?
Regarding the Confluence of AI: In the post, you said that “integration with AI Agents” will be a future project. What do you consider to be the most promising initial uses of an AI agent in this context, considering that your architecture effectively surfaces structured on-chain events? Would it be more useful for natural language data querying, operational automation, or predictive analytics?
Fantastic content; once again, I appreciate you compiling and sharing it! I hope you intend to share more of this adventure; I will surely be watching for more! 
Thanks for the feedback and queries! Yeah I think its good to have more of these kind of reproducible showcases, particularly ones which show and prove the modularity and extendibility.
I’d say extending with custom on-chain trackable data, currently is quite an obstacle for businesses without dedicated teams. It does take some domain modelling and overriding of quite a bit of ingestion and publishing logic. Have had discussion with the CF team, its on their roadmap to make this a bit simpler, but would still require custom dev for custom data. So that is a barrier for sure. I could imagine this as no-code or ontology based that a domain expert could define and be auto compatible with Reeve, would need a nice ui/platform though. Another large barrier for general purpose use (other than as-is usage of financial reporting intended for public accountability) is of course still needing privacy and probably deployment on layer2 so costs are competitive. However, if the goal is to give a business ERP capabilities at the same time as using Reeve in its current form, its not that big of an obstacle. The first step as you say, organising their messy data, I believe could be done in the Netsuite cloud based tool? I haven’t used it myself, however imagine this is no different an obstacle for businesses as adopting other new IT tools and there should be training available. Then, since Reeve is integrated to it - for syncing intentionally public financial data - there isn’t any need for custom domain modelling in Reeve itself so the barrier is low. Therefore, different barriers for different purposes.
I imagine for the as-is scenario its just Netsuite training, and basic maintenance training for a Reeve server (cnode + Reeve webapp) - e.g. along the lines of an IT support person turning it on and off. n.b. the webapp front-end may still need to be created custom, not exactly sure what CF offer as hosted solution or what extent can re-use their demo front-end. Not sure if this answers your question.
YaciStore is good, extendable, event based. It’s data arch is a bit bloated imo, not as normalised as DbSync or others / can’t do fancy queries with as good performance, but its good for what it is and will only get better. Built using very flexible and well supported Spring framework. I had to modify it to get my demo working, however preferred not to. It wasn’t particularly difficult, just bit more complexity to manage. Only needed one minor extra property for my case, and likely only because I was explicitly trying to match their design pattern. This was another discussion I had with CF, there are probably ways they can avoid this; e.g. pub-sub events over to Reeve, or even just avoiding needing any custom indexing at all, relying on properly indexed metadata queries, tbd. You could definitely and quite simply modify it if needed, to get those extra event details you mentioned.
Mainly wanted to indicate desire to build and investigate a big graph engine with data fusion / ingestion pipelines, shared ontologies, as an exploration of that tech including graph RAG and other ways to augment AI. To answer your question though, trusted autonomy would be a big utility, having the most highly trusted data layer on the planet to get ground truth from. Although this has other challenges such as how to ensure veracity/truthiness of the data content.
Those were hard questions, I’ll send an invoice to CF for the time!
Agree, need to get the adoption of these kind of use-cases in full-gear. Atm I think support / consulting is needed to help everyone from small to medium to large enterprises get up and running. It should get easier over time, but still needs at minimum some cloud compute to run cnode, servers and custom apps, in order to benefit from the espoused benefits of trust and inclusive accountability. This is the kind of work I’m looking for, if anyone listening has any options! Do intend to share more adventures at some point, cheers!