iiinova — where your digital self keeps evolving

Ron Itelman @ronitelman.bsky.social · 26/09/2026

Excited to be learning about jev, a new and needed paradigm is continuing to gain momentum. To be clear, were following similar strategies (and have been before jev) with the W3C Context Graph Community Group

010

Ron Itelman @ronitelman.bsky.social · 22/08/2026

Without a shared definition of context it is extremely difficult to manage it across silos.

000

Ron Itelman @ronitelman.bsky.social · 18/08/2026

Much of the industry is now focused on providing more context to models: larger context windows, RAG, memory, and better prompting. These are important. The category error is treating them as the context problem without first making "context" itself computationally measurable.

000

Ron Itelman @ronitelman.bsky.social · 16/08/2026

What is context? Can you answer what is in cell d8? Decide before you read on... The answer is that this is a map of the sky and d8 refers to a different map. That missing information, revealed to you which changed your understanding, is context.

000

Ron Itelman @ronitelman.bsky.social · 05/08/2026

Major breakthrough! By using the four facets of the context graph, was able to demonstrate the gauge can predict which missing information to ignore to safely make decisions. Calling it decision shapes for now.

000

Ron Itelman @ronitelman.bsky.social · 18/07/2026

Context saves money because time, effort, and tokens aren’t wasted in avoidable misunderstandings. Context is the process of identifying what information is missing, for making better decisions and coordinating correctly. This is what we're working on in the W3C Context Graph Community Group.

010

Ron Itelman @ronitelman.bsky.social · 08/07/2026

Unifying Business, Data, and Code is getting translated to Chinese!

000

Ron Itelman @ronitelman.bsky.social · 04/07/2026

The Four facet model for AI data product observability is simple and effective.

000

Ron Itelman @ronitelman.bsky.social · 28/06/2026

Unless you can explain how your benchmark handles ambiguity, using agents to merge datasets, and touting your own benchmarks aren't ready for others to trust.

000

Ron Itelman @ronitelman.bsky.social · 18/06/2026

The circular relationships, dependencies, and expectations of trillions in revenue, with plenty of evidence to the contrary makes it hard to fathom exuberance in markets. Feels like dominoes are about to fall...

010

Ron Itelman @ronitelman.bsky.social · 17/06/2026

When people ask me how to get started in R&D, innovation, or skunkworks, I do my best to help. I am saddened that leaders are willing to take high risk bets in AI while employees get little to no support for their ideas to experimented on with little risk to them.

000

Ron Itelman @ronitelman.bsky.social · 16/06/2026

Frame: an agent's way of reading messages: the interpretive setting a message is read against. It lives with the agent, not in the message. Framing: the act of supplying a receiver the frame a message did not carry ("let me give you context"), changing how the receiver reads what it already has.

000

Ron Itelman @ronitelman.bsky.social · 15/06/2026

The four facet model is the foundation of the w3c context graph community group protocol. Without making these four facets explicit, you introduce dark uncertainty which no policy of a model can manage, specifically with communication between nodes in a network. It is generalizable and scale-free.

010

Ron Itelman @ronitelman.bsky.social · 09/06/2026

Forward Deployed Engineers and ontologies are hot. But before you buy an AI solution: • How is accuracy measured? • How is complexity minimized? • How hard is it to leave? Building an open standard & looking for pilot partners to benchmark the above with the W3C Context Graph CG spec.

010

Ron Itelman @ronitelman.bsky.social · 01/06/2026

Would you like to learn about intent mapping for the W3C Context Graph CG spec? DM me!

000

Ron Itelman @ronitelman.bsky.social · 24/05/2026

Leaders can mistakenly assume alignment exists because strategy has been communicated. But in reality, people align to individual incentives, budgets, and KPIs long before they align to organizational goals. Systems can appear stable in isolation while failing in coordination.

The image uses the metaphor of an ostrich burying its head underground to represent organizations and leaders intentionally ignoring systemic problems that fall outside local incentives or KPIs. The title, “Choosing Darkness,” frames this not as accidental blindness, but as a structural decision driven by misaligned incentives.

The ostrich’s body remains visible and exposed while its head disappears into a dark hole, symbolizing teams or decision-makers who avoid seeing operational realities that are inconvenient, costly, or not directly tied to their measured success. The subtitle, “Incentives for one, Success for none,” reinforces the central critique: organizations often reward local optimization even when it damages the broader system.

The image represents how leadership can mistakenly assume alignment exists because strategy has been communicated, while teams actually align to incentives, budgets, and KPIs. In these environments, cross-functional risks become invisible because no single group is rewarded for owning them. Problems that threaten the organization as a whole are ignored if resolving them creates local cost without local benefit.

The visual critiques incentive structures that prioritize departmental performance over coordination, transparency, and shared operational responsibility. Its deeper message is that systemic failure in organizations often emerges not from malice or incompetence, but from rational actors optimizing within fragmented measurement systems.

Illustration by Ron Itelman, W3C Context Graph Community Group Chair, May 2026

030

Ron Itelman @ronitelman.bsky.social · 10/05/2026

Has lean solves waterfall issues? Will AI solve ______ issues? Not if those issues: Mainly coordination in complex networks, made explicit and testable, are not addressed.

000

Ron Itelman @ronitelman.bsky.social · 08/05/2026

What we're working on, in the w3c context graph community group...

000

Ron Itelman @ronitelman.bsky.social · 07/05/2026

The reason AI projects struggle with implementation is that same that has been a challenge for all software development. We have different reference frames, which give us valid reasons to make decisions within those frames. But an organization doesn't live in one frame. Those differences? Context.

000

Ron Itelman @ronitelman.bsky.social · 15/04/2026

@stevenstrogatz.com 10 years ago your TED talk on sync changed how I think. It led me to a new theorem on unverifiable configurations at system boundaries. The unexpected finding: some errors aren't random — they phase-lock to the calendar (date format ambiguity fires on days 1–12 of every month)

020

Ron Itelman @ronitelman.bsky.social · 08/03/2026

Schrodinger's revenue: Two columns, one word

000

Ron Itelman @ronitelman.bsky.social · 07/03/2026

Everyone claims to own semantics — ontologists, knowledge graph engineers, Decision Intelligence. They're all right, inside their own systems. The W3C Context Graph CG formalizes the liquid space between them — the boundary no one owns.

010

Ron Itelman @ronitelman.bsky.social · 06/03/2026

An imaginary Stephen King novel...Trust Us “At first, the systems worked perfectly. AI doctors caught diseases earlier than any human could. AI lawyers won cases in minutes. AI agents negotiated contracts, managed cities, ran the courts. ...

100

Ron Itelman @ronitelman.bsky.social · 03/03/2026

You depend on systems you can't see inside: "Revenue" in one means gross. In another, net. If nobody checks if intent is coherent, a blast radius risk silently grows. The W3C Context Graph Community Group is building a protocol to tell you what's aligned, what's unknown, and what to do.

Illustration by Ron Itelman, Chair of the W3C Context Graph Community Group, showing a stick-figure Gandalf character standing inside a glowing protective bubble, holding a staff and sword, facing a large dark Balrog creature with wings and fiery eyes looming behind. Text reads "You Shall Not Pass!" The bubble represents a local context protocol protecting against the chaos of uncontrolled external systems.

000

Ron Itelman @ronitelman.bsky.social · 05/01/2026

Decision traces are about modeling a user's understanding in decision making...

010

Ron Itelman @ronitelman.bsky.social · 03/01/2026

Exploring concepts of confidence, conviction, and trust from a product design perspective...

010

Ron Itelman @ronitelman.bsky.social · 02/01/2026

Working on context graphs and organizational intelligence? I've got tools, methodology, and experience to help, please reach out!

000

Ron Itelman @ronitelman.bsky.social · 29/12/2025

There's a lot of buzz around context graphs and I want to introduce the idea of intent maps. Intent Maps are real time and inherently feedback loops of the user defined meaning for diff operations against ontologies and system rules. I've opened sourced core libraries, please reach out!

000

Ron Itelman @ronitelman.bsky.social · 28/12/2025

A few thoughts on the need to define context a bit more formally: I wanted a way to compute relative shifts of observer knowledge that is compatible with tensor logic as I research neurosymbolic AI. Still working through understanding the math, but here's the simple principle...

100

Ron Itelman @ronitelman.bsky.social · 14/12/2025

My belief is that the intent management layer is missing from most companies attempts to build reliable AI solutions.

000

Ron Itelman @ronitelman.bsky.social · 25/11/2025

Released trustquery language (open source) for real-time metadata standards to minimize uncertainty in conversations about data. trustquery.com

000

Ron Itelman @ronitelman.bsky.social · 21/11/2025

GenAI for finance has serious flaws that without guardrails make it irresponsible for companies to adopt. Latest research I worked on showed ambiguity in a query for ONE word would swing answers by millions, and the AI model's ability to manage the risk was zero. Guardrails BEFORE hitting submit!

000

Ron Itelman @ronitelman.bsky.social · 18/11/2025

Here's how trustquery.com works for tabular data (like a .CSV): 1. As you type (or paste) your data into the chat, ambiguous columns are detected. 2. It notifies the user and asks a clarifying question 3. Any user feedback is added as metadata directly into the column name. Open source (MIT).

010

Ron Itelman @ronitelman.bsky.social · 18/11/2025

Released open source library to turn any phrase into an interactive element: menus, api calls, guardrails, etc. I'll be releasing more open source libraries on TrustQuery.com focused on pre-processing info for higher accuracy AI experiences. Please reach out if you'd like to discuss! Ron

110

Ron Itelman @ronitelman.bsky.social · 10/09/2025

This is not going to end well, and if your company's employees are putting valuable IP into their systems, what happens to the data if one of your competitors (or their IT suppliers/partners) buys a bankrupt OpenAI and gets a treasure trove of data?

000

Ron Itelman @ronitelman.bsky.social · 20/08/2025

If your business depends on A) an unprofitable business that B) needs to spend trillions of dollars, I have a question for you: What happens to your business if your AI provider suddenly collapses?

000

Ron Itelman @ronitelman.bsky.social · 22/07/2025

Saw this excellent talk on OpenAI's intent modeling approach. Glad to see Intent Engineering and ambiguity detection and correction part of the AI conversation. This is exactly what TrustLoop is working on: aligning intent across business, data, and UX layers for natural-language-query to sql.

010

Ron Itelman @ronitelman.bsky.social · 25/06/2025

I've launched a new company, TrustLoop.AI, to eliminate the main source of hallucinations in AI, and my client said this phrase and it's only better than any marketing I could have come up with. I'm not sure how much I'll owe him for it 😂

010

Ron Itelman @ronitelman.bsky.social · 27/05/2025

Working on private MCP Registries for companies working on AI & MCP, please reach out if you are available to chat, I'd love to hear about your core challenges.

020

Ron Itelman @ronitelman.bsky.social · 14/05/2025

The biggest problem with "Vibe Coding" - identifying hacks so that you aren't leaving your systems vulnerable.

020

Ron Itelman @ronitelman.bsky.social · 11/04/2025

Initial results: → 7.2 out of 25 cells contain errors (28.8%) → Average error magnitude in wrong cells: 8.4% After just 5 interactions, nearly 1/3 of your critical financial data could be corrupted.

010

Ron Itelman @ronitelman.bsky.social · 11/04/2025

using three different LLMs (as examples of agent systems cooperating), one model had one cell wrong, and also verified it incorrectly.

110

Ron Itelman @ronitelman.bsky.social · 11/04/2025

Using a small and public dataset (FINRA bond activity)....

110