ETL to QE, Update 57, Roadmap Animation

Date: 2025-01-06 and 2025-01-10

As yet another reminder, Greatness is Other People, and this little DDaemon 2025 Roadmap Rev. 0.0.1 does not involve any other people in it. Not involving other people violates the User Centered Design principals outlined in my personal design bible

When trying to follow through on step 1, Nostr interface equivalent to Open WebUI I came across the meme schema problem from step 5, Knowledge Graph all the Things. Which was missing the RBAC requirement that's outlined in steps 2, Nostr RBAC LDAP via Blossom Upgrade with Multiformats IPFS functionality and Basic Agents Nostr tokenized RBAC with wiki publishing. So seems like we may have to change up our priorities.

The Nostr interface equivalent to Open WebUI sprint was supposed to be easy. SKOP TO NEXT SECTION FOR RESULTS

Allow users to tag the bot in threads and have it reply, simple.
Be able to interact with the bot via encrypted direct messages, easy.
Slash Commands to specify LLM settings, A custom parser should simple enough, there also doesn't exist any NIP or Nostr UI that supports Slash Command helpers. We also lose the Provenance functionality we want from step 6. Agents and ETL Kafka Hamilton DAG style. Also there is no NIP standard that makes adding the slash command tags to events easy, which will also require a custom client.
- Maybe it is better to specify user settings via separate conversations and have the bot fetch latest settings event when responding to a user. But that is A LOT of additional complexity and leads to the application no longer being stateless.
Publish private LLM conversation for others to see, I see implementation difficulties. Doable with Slash Commands, the bot will just post the entire convo with placeholders for the user. We will also need a way for users to tell the bot what to delete.
CUSTOM CLIENT FEATURES
Apply editable title / description to each LLM conversation that is supposed to be encrypted, Nostr doesn't natively do this without repurposing the correct NIP or NIPs. This is something that will require a custom client.
LLM Text Generation Streaming, we will need a custom NIP for that, and it will only work in our specific client. That requires an entire individual POC.
Setting System, Bot, User author tag for messages.
Editing the response of the bot.
Generating the Text Graph of the conversation.
Rating text that's generated, we can use NIP07
Adding Knowledge Graph Tags

Schema Comes Before Application

As a reminder this is the OG roadmap,

See DDaemon 2025 Roadmap Rev. 0.0.2

The Slash Command Nostr LLM Bot was a good starting point but doing the Nostr Open Web UI Interface was not the correct move.

But the roadmap should look like,

So our priority is a BOT framework rather than a Open WebUI rip off running on Nostr. There are a couple checkpoints we need to reach for this to be successful.

Research Document and published blog post on Other Bot Systems such as Mastodon, Discord, Telegram etc. etc.
Research Document and published blog post on Existing Nostr Bots
A script to Scrape Bot Interactions
The best documented Nostr Bot Framework
The Nostr Bot Framework should work with NATS for when we get to Agents and ETL Kafka Hamilton DAG style
Do an experiment where I boost my own posts with Sats and see if it helps with Engagement at all
For the bot we also need to be able to track what Relays we were sucesfully able to publish our messages to
We also need to worry about spamming
We need to interview people about the Bots they use and like
We need to find the most popular bots out there

Ah there we go, actually fletching out the milestones with tasks that can be measures with yes or no answers rather than making 1-3 sentence aspirational notes.

It is time we took the Demo Pill and started living by the values I actually believe in.

Backlinks

Project Update Posts