RDF-backed collection

The Context Gap

Jessica Talisman's critique of context graphs as the market rediscovery of process knowledge, expanded with the visible LinkedIn comment thread.

Overview

The article frames the enterprise AI context gap as a failure to preserve decision traces, procedural knowledge, institutional memory, and the human infrastructure of knowledge management. The comment thread extends the argument into public decision making, EU AI Act compliance, observability, Linked Data publication, and the cultural work required behind any graph artifact.

Core Claims

A New Name for an Old Wound

A New Name for an Old Wound In December 2025, Foundation Capital published an essay titled 'AI's Trillion-Dollar Opportunity: Context Graphs,' which has since become one of the most-discussed pieces in enterprise AI. The thesis holds that agents are hitting a wall that governance alone cannot solve, and the wall is not missing data - it is missing decision traces. Foundation Capital proposes the 'context graph' as the answer, characterized as a living record of decision traces stitched across entities and time, whe...

Outsourcing and Process Knowledge Loss

What Foundation Capital calls a context graph is, in the disciplinary vocabulary of information science, a procedural knowledge graph grounded in formal ontologies, with provenance, temporal validity, entity resolution and controlled vocabularies underneath. We already have the methods. PROV-O models provenance. SKOS handles vocabulary control. OWL provides the reasoning structures. The toolkit predates the marketing. What is missing is not tooling but the recognition that this is a knowledge management problem, re...

The Great Unbundling

When we sent manufacturing to China, India and the Philippines, we divested from the opportunity to learn, iterate, fail and improve. We eliminated critical feedback loops, a requisite for capturing and documenting procedural knowledge. We dissolved communities of practice that essential sources for process knowledge. And crucially, thanks to gapping holes in the end-to-end process knowledge fabric, we stopped investing in the knowledge infrastructure required to capture and maintain our understanding of how comple...

Shenzhen and Process Knowledge

What went largely unexamined was what happened to process knowledge when these activities migrated. The assumption was that process knowledge could be cleanly separated from execution - that 'knowing how' could remain in Western headquarters while 'doing what' happened elsewhere. This assumption proved catastrophically wrong. Look no further than current struggles in developing knowledge infrastructures in technology organizations and massive failures of agentic AI systems. (see my series, 'Why AI Isn't Autonomous ...

Visible Comment Thread

Captured from the LinkedIn article aside after expanding visible comment text. The page now reports 10 visible comments, including top-level comments and author replies.

Kingsley Uyi Idehen
Founder & CEO at OpenLink Software | Driving GenAI-Based AI Agents | Harmonizing Disparate Data Spaces
1m

Jessica, here is a knowledge graph, deployed as a Semantic Web, constructed from notes generated by my AI agent from this post. Publication for Web access and upload to the underlying Virtuoso-hosted knowledge graph is achieved by mounting a Virtuoso WebDAV folder to the local operating system, copying local files into the mounted folder, and letting Linked Data principles provide Web-scale data access by reference. Links include https://linkeddata.uriburner.com/DAV/demos/daas/context-gap-jessica-talisman-gpt5-1.html and hashtags #AgenticWeb #ContextGraph #KnowledgeGraphs #LinkedData #SemanticWeb #HowTo #UseCase.

RDF entity
Charles Ivie
Partner, Data & AI Engineering Lead | ex AWS, BBC | Ontologist
3h

Very astute observations Jessica Talisman. It reminds me of so many decisions that are made by every government. Quick decisions are made with unknown consequences yet to be realised. Often resulting in the cure being worse than the original pathogen. We are, as a species, so destined to find a quick fix, only to realise it cases a slow decline.

RDF entity
Jessica Talisman
Author, Building information systems for the benefit of all Taxonomy | Ontology | Knowledge Graphs
3h

Charles outsource or automate without much in between. Currently reading Goliath's Curse by Luke Kemp which I think you'd appreciate

RDF entity
Fred Lardaro
Let's make your home more safe and secure with a self-monitored video sensor solution that provides one-touch 911.
58m

Jessica Talisman!!! Everything old is new again. If all our experts disappeared... would our decision making systems be able to explain how decisions were made? One consolation is that we are pretty good at gaining wisdom after the fact.

RDF entity
Jessica Talisman
Author, Building information systems for the benefit of all Taxonomy | Ontology | Knowledge Graphs
55m

Fred knowledge and wisdom!

RDF entity
Andrei Bâcu
Founder @Antifragile.AI | Senior Architect | GenAI | Data | AI/ML @ Flutter Entertainment
2h

Jessica Talisman I really appreciate that you are sharing all this valuable information in an usable format. Have started building a knowledge graph and an ontology for the EU AI Act at Antifragile.AI, for streamlining compliance & cybersecurity and helping systems to become antifragile in time. Context graphs and context engineering are core components for providing evidence-based assertions and remediations, advanced observability and even inputs for RL of Agents.

RDF entity
Andreas Schurch
Founder, Duodata | I help data leaders go from messy KPIs to a business-approved metrics layer everyone trusts | Ex-AWS, Microsoft, Matillion
2h

Jessica, love this! The phrase "context graph" is useful, but your point that the market is rediscovering process knowledge, ontology, and institutional memory is spot on. Curious how others here are thinking about the cultural side of this: can companies rebuild the communities of practice behind the graph, or will they keep trying to buy the artifact without restoring the discipline? We have seen this movie before.

RDF entity
Andi Willmott
Chief Solutions Officer
2h

What is lost is found again - just repackaged. Jessica Talisman, really enjoyed your observations linking the past days to today - hugely astute.

RDF entity
Jessica Talisman
Author, Building information systems for the benefit of all Taxonomy | Ontology | Knowledge Graphs
2h

thank you, Andi

RDF entity
Shashi Bhatnagar
Planetary Risk managed, Spatially Intelligent, Adaptive Resilient Assets & Solutions.
15h

This is brilliant articulation of risks we face - and what to do to influence the mitigation. A very multidimensional view can be formed from the same article.

RDF entity

References

Books, essays, standards, footnote sources, and public identity links named or linked by the source article and visible comment thread. High-confidence DBpedia and Wikidata cross-references are modeled with owl:sameAs.

Essays and Articles

Context Graphs, One Month In

Follow-up Foundation Capital essay referenced for Dharmesh Shah and the system-of-record-for-decisions framing.

Author/source: Foundation Capital

Source

Context Data Platform

Glean article referenced for Arvind Jain and the claim that the context graph concept finally has a name.

Author/source: Glean

Source

Why AI Isn't Autonomous Yet

Related Substack series linked in the article when discussing knowledge infrastructure failures and agentic AI systems.

Author/source: Jessica Talisman

Source

Studies and Source Notes

Process Knowledge Management

Substack source linked through footnotes for KPO, market estimates, and the process-knowledge-management continuation.

Author/source: Jessica Talisman

Source

Books

Standards and Specifications

External Identity Links

Source Media Embeds

The source LinkedIn document includes publishing embeds in and after the podcasts section. They are modeled as schema:VideoObject resources with schema:embedUrl and linked into the graph.

Knowledge Graph Explorer

Advanced KG Settings
Predicate Filters
Zoom active. Click outside to release.

FAQ

What is the context gap?

The context gap is the difference between knowing what happened and knowing why a decision was justified, permitted, and executed in a given operational moment.

Why does the article critique the phrase context graph?

The article argues that context graph is a market-friendly name for process knowledge, procedural knowledge, institutional memory, and formal knowledge-management practices that already existed.

What role did outsourcing play in the argument?

Outsourcing is presented as a long-term driver of process-knowledge loss because organizations separated execution from the environments where knowledge was generated, documented, and transmitted.

How do comments extend the article?

The comments connect the thesis to government decision making, wisdom after the fact, EU AI Act compliance, context engineering, observability, Linked Data publication workflows, and the need to rebuild communities of practice.

What is missing if companies only buy context graph software?

Software alone cannot restore documentation practices, apprenticeship lineages, communities of practice, or cultural respect for knowledge engineering work.

Which semantic-web standards does the article mention?

The article mentions PROV-O for provenance, SKOS for vocabulary control, and OWL for reasoning structures.

How is process mining used as an analogy?

Process mining is used as an analog because companies such as Celonis built businesses by helping enterprises see workflows they had ceased to understand.

What is a procedural knowledge graph?

A procedural knowledge graph is an ontology-grounded graph that captures process knowledge with provenance, temporal validity, entity resolution, controlled vocabularies, and decision traces.

Why are communities of practice central?

They are the social infrastructure through which operational knowledge, craft judgement, documentation habits, and apprenticeship practices are sustained.

What does the comment thread add about culture?

The thread asks whether companies can rebuild the communities of practice behind the graph or will keep trying to buy the artifact without restoring the discipline, while also showing how practitioners can publish and inspect graph artifacts directly.

Glossary

Context graph

A living record of decision traces stitched across entities and time, making precedent searchable.

Context gap

The gap between recorded facts and the reasoning, permissions, and conditions that made decisions possible.

Decision trace

Evidence of why a decision was allowed to happen at a particular time.

Process knowledge

Operational know-how about how work is performed, improved, and transmitted.

Procedural knowledge graph

A graph grounded in formal ontologies, provenance, temporal validity, entity resolution, and controlled vocabularies.

Institutional memory

Organizational knowledge accumulated through documentation, apprenticeship, and communities of practice.

Knowledge management

Discipline concerned with capturing, organizing, maintaining, and applying organizational knowledge.

Community of practice

Social structure through which practitioners transmit process knowledge and improve work.

Outsourcing

Transfer of work outside an organization, treated in the article as a driver of lost process knowledge.

Knowledge Process Outsourcing

Outsourcing of specialized knowledge-intensive work such as legal research, analytics, engineering design, and research.

Business Process Outsourcing

Outsourcing of routine transactional processes such as call centers and data entry.

Process mining

Method for analyzing event logs to reveal operational workflows; used as an analogy for context graphs.

PROV-O

W3C provenance ontology for modeling provenance information.

SKOS

W3C model for controlled vocabularies, concept schemes, and thesauri.

OWL

Web Ontology Language used for formal reasoning structures.

EU AI Act

Regulatory context mentioned in the comments as a target for knowledge graph and ontology work.

Context engineering

Practice of supplying agents and systems with evidence-rich context.

Agentic AI systems

AI systems whose failures are framed as symptoms of weak process knowledge and missing decision traces.

Semantic Web

Web of linked RDF data and dereferenceable identifiers used to publish knowledge graphs for machine and human access.

Linked Data

Principles for publishing structured data on the Web using HTTP IRIs and typed links between resources.

WebDAV

HTTP extension used here as a mounted file folder workflow for publishing generated RDF and HTML artifacts.

Virtuoso WebDAV publication

Workflow where files copied to a mounted Virtuoso WebDAV folder become Web-accessible resources and graph inputs.

Risk mitigation

Comment-thread framing that the article can inform mitigation of multidimensional organizational and AI risks.

Multidimensional view

Comment-thread idea that the same article can support many perspectives when represented as a graph.