Re: The Semantic Web Vision is finally ready to hit escape Velocity!

Hi Michael,

On 9/26/26 10:22 AM, Michael Brunnbauer wrote:
> Hello Kingsley,
>
> On Fri, Sep 25, 2026 at 07:22:18PM -0400, Kingsley Idehen wrote:
>> Naturally, there's a live demo
> Which does not work.

Ah! I just realized the KG explore component isn't working. Please note, 
every link in the page resolves to a loosely coupled RDF graph (the 
basic live demo). The KG explore visualization in this particular post 
is now fixed.


Note, https://linkeddata.uriburner.com/weblog/ is full of live demos in 
every single post, and the KG explorer components are working too.

Also note the fundamental demo, which slipped out of the blog post:
https://muse.ai/s/northwindkg-reactive-dashboard-xsl5xyxlkoxfxqsc

That's showing a conventional relational DBMS in Semantic Web form, 
courtesy of an RDF view layer.

> Personally, I take this as a good omen.
>
>> *In my opinion, the arduous journey getting The Semantic Web Project to
>> escape velocity is over.*
> It will never come back? :-)

My point isn't about coming back. It's actually about moving forward now 
that a complementary client layer, in the form of AI Agents, lies in wait.

The fundamental point has always been:

Use hyperlinks to name entities and relationships. Do that, and you end 
up with a webby entity relationship graph. That builds on the broadly 
using document network.

Remember:
A network is a graph comprising nodes of the same type connected by 
common edges (i.e., edges of a specific type) . In RDF parlance that 
means: a graph where the subjects and objects are of the same type 
connected by the same predicate e.g., a graph comprising  entities of 
type "document" where each document connects to another via a "links-to" 
predicate.

>
>> Today's LLM-powered AI Agents have been trained on all the key
>> specifications, dramatically reducing implementation and usability friction
>> that held things back for decades.
> It's dangerous to anthropomorphize LLM-powered AI - an alien intelligence -
> but it's also very useful to quickly communicate views and points. So I'm gonna
> do it here:
>
> I don't think the availability of cheap labor is what held things back for
> decades.
>
> That this labor can be characterized as artificial sociopaths with savant
> syndrome poses additional problems of trust outside "niche" applications like
> making harmless demos, generating formally verifiable work or explaining things
> to students who will have their understanding tested without AI help.
>
> "Never trust AI output" is my current credo.

I am not an advocate of anthropomorphic marketing of langulators (what 
LLMs are). My point about LLM and RDF symbiosis, which is more to do 
with neuro-symbolic AI and in no way supporting that kind of nonsense.

I too believe in: never trust, always verify. That's why I emphasize 
"harnessing" when working with langulators.

>
>> Simply tell your Agent (*a context rich harness wrapped around your
>> preferred LLM*), in natural language, what you want done with the data???and
>> what kind of artifact you want to visualize the results.
> Now that seems like a sufficiently bounded example. The data is already there.
> Getting, understanding and massaging that data was always the main problem.

Yes.

> The agent only generates a read query plus visualization, which is discarded
> afterwards.
>
> But this is not specific to the Semantic Web and not what held it back.

Again, the Semantic Web Project is simply about using naming conventions 
to creates a Web where entity relationships are informed by logic 
expressed in an ontology.

>
> I can see how trusting the agent is less important here, but I don't think
> it is irrelevant.
>
>> The standards were already there.
>> The infrastructure was already there.
>> The data was already there.
> The last statement does the most work here, hand waving away the real problem -
> if you ask me.

I don't hand wave, which is why I spend a lot of time working on 
showcase examples before sharing anything in public :)

BTW -- here's an alternative edition of the post:
https://www.linkedin.com/pulse/semantic-web-vision-escape-velocity-via-ai-agents-kingsley-uyi-idehen-bebhc/

Kingsley

> Regards,
>
> Michael Brunnbauer
>


-- 
Regards,

Kingsley Idehen 
Founder & CEO
OpenLink Software
Home Page:http://www.openlinksw.com
Community Support:https://community.openlinksw.com

Social Media:
LinkedIn:http://www.linkedin.com/in/kidehen
Twitter :https://twitter.com/kidehen

Weblogs:
Bleeding Edge AI and Semantic Web Stuff:http://linkeddata.uriburner.com/weblog/
Company:http://www.openlinksw.com/weblog/