Executive Summary
The blank page is the most expensive moment in knowledge work.
It is the moment when a lawyer opens a new document to draft a brief and stares at the cursor. The moment when a consultant opens a slide deck for a new client presentation. The moment when a developer opens an empty file to architect a new system. The moment when a writer begins a new article.
The blank page is not actually blank. The professional has relevant knowledge -- from previous briefs, previous presentations, previous architectures, previous articles. But that knowledge is scattered across old files, old emails, old conversations, and old memories. Assembling it into a starting point takes time -- often more time than the actual drafting.
This is the cold start problem, and it costs knowledge workers an estimated 25-40% of their productive time. Not because they lack knowledge, but because their knowledge is not accessible at the moment they need it.
Kent eliminates the blank page. When you start a new task, Kent's knowledge graph already contains your relevant previous work, your preferences, your domain expertise, and your established patterns. The page is never blank. It is pre-loaded with everything you know.
1. The Cost of Starting From Zero
1.1 The Cold Start Tax
Microsoft's 2025 Work Trend Index found that knowledge workers spend an average of 1.8 hours per day searching for information they need to begin or continue their work (Microsoft, 2025). This includes searching email for previous correspondence, finding old documents for reference, locating meeting notes from related discussions, and reconstructing context from memory.
At a median knowledge worker salary of $78,000/year, 1.8 hours/day of search time represents approximately $18,000/year per employee spent not on productive work, but on finding the information needed to begin productive work.
For a 50-person knowledge-work team, the cold start tax is approximately $900,000/year -- nearly a million dollars spent on the friction between having knowledge and accessing it.
1.2 The Scatter Problem
The cold start tax exists because relevant knowledge is scattered across an average of 7-12 different applications (McKinsey, 2025):
- Previous work product (Google Drive, SharePoint, local files)
- Correspondence (Gmail, Outlook)
- Meeting notes (Notion, OneNote, handwritten)
- Project management (Asana, Jira, Linear)
- Messaging (Slack, Teams)
- Calendar (for timeline context)
- Browser bookmarks and history
- Memory (the least reliable source)
No single search covers all of these. Google Drive search finds Drive documents but not emails. Outlook search finds emails but not Notion pages. Slack search finds messages but not files. The professional must run multiple searches across multiple applications, mentally aggregate the results, and construct a starting context from fragments.
This is not a search problem. Search within each application works well enough. It is an aggregation problem -- the inability to search across all sources simultaneously and receive a unified result.
1.3 The Expertise Paradox
The professionals who suffer most from the cold start problem are the most experienced ones. A junior employee with two years of history has a small body of previous work to search. A senior professional with fifteen years of history has a vast body of relevant experience -- but it is scattered across a decade of emails, documents, presentations, and conversations that no search tool can effectively traverse.
Experience should be an accelerator. Instead, it often becomes a brake -- the more you know, the longer it takes to find what you know, and the more likely you are to start from scratch rather than endure the search.
2. The Pre-Loaded Page
2.1 How Kent Eliminates the Blank Page
When a Kent user begins a new task, the knowledge graph already contains the relevant context. The mechanism is straightforward:
- The user highlights text related to their new task (an email, a brief, a project description) and triggers Kent
- Kent's context builder queries the local knowledge graph using embedding similarity to find related nodes
- Related nodes -- from previous conversations, ingested documents, connector queries, and past skill executions -- are assembled into context
- The AI receives both the current task and the relevant historical context
- The response is informed by everything the user has previously worked on that is related to the current task
The page is never blank because the knowledge graph is always populated. The user's entire professional history -- or at least the portion that has been ingested and connected -- is available as context for every new task.
2.2 Context That Finds You
The critical difference between Kent's approach and traditional search is directionality. Traditional search requires the user to formulate a query, identify which application to search, and evaluate results. The user must know what they are looking for before they can find it.
Kent's context builder reverses this. The user does not search for relevant context. The context finds the user. When the knowledge graph is queried with the current task's embedding, semantically related nodes surface automatically -- including nodes the user may have forgotten existed.
A lawyer drafting a new contract clause may not remember that they handled a similar clause for a different client eight months ago. But the knowledge graph remembers. The embedding similarity between the current task and the historical work surfaces the relevant precedent without the lawyer needing to remember it, search for it, or even know that it exists.
2.3 The Compound Context Effect
The pre-loaded page gets richer over time. A Kent user in their first month has a sparse knowledge graph -- the blank page is less blank, but not dramatically so. A Kent user in their twelfth month has a dense graph with thousands of interconnected nodes. The blank page is densely pre-loaded with relevant context from a full year of professional activity.
This is the compound context effect: the longer you use Kent, the less blank the blank page becomes, and the faster you can begin productive work on any new task.
3. Use Cases: The Blank Page Eliminated
3.1 Legal: Brief Drafting
Traditional workflow: Lawyer receives a new case. Spends 2-3 hours searching for similar cases they have handled, locating relevant precedent from their firm's document management system, finding correspondence with the client, and assembling a starting framework for the brief.
Kent workflow: Lawyer highlights the case summary. Kent's knowledge graph surfaces relevant nodes: previous briefs on similar topics, client correspondence history, relevant statutes discussed in past conversations, and the lawyer's preferred brief structure. The AI generates a first draft that incorporates all of this context. The lawyer edits and refines rather than writing from scratch.
Time saved: approximately 2 hours per new brief, based on the difference between the search-and-assemble workflow and the refine-from-draft workflow.
3.2 Consulting: Client Presentations
Traditional workflow: Consultant starts a new engagement. Spends 3-4 hours reviewing previous engagements in the same industry, finding relevant frameworks and methodologies, locating market data from past research, and building a presentation structure.
Kent workflow: Consultant describes the new engagement to Kent. The knowledge graph surfaces relevant nodes: industry analysis from previous engagements, applicable frameworks, market data, and the consultant's preferred presentation structure. The AI generates a structured outline with relevant content pre-populated.
Time saved: approximately 3 hours per new engagement kickoff.
3.3 Software Development: Architecture Design
Traditional workflow: Developer begins a new system design. Reviews previous architectures for similar systems, searches for relevant design patterns, locates API documentation for planned integrations, and recalls lessons learned from past projects.
Kent workflow: Developer describes the system requirements. Kent surfaces nodes from previous architecture decisions, relevant design patterns discussed in past conversations, integration documentation previously ingested, and lessons learned from similar projects. The AI generates an architecture proposal that incorporates institutional knowledge.
Time saved: approximately 1-2 hours per new system design.
4. The Architecture of Readiness
4.1 Continuous Ingestion
The pre-loaded page requires continuous ingestion -- a steady flow of professional activity into the knowledge graph. Kent achieves this through three ingestion paths:
Passive ingestion. Connectors to Gmail, Google Drive, Notion, and other tools automatically ingest new content as it appears. The user does not need to manually import anything -- the knowledge graph grows as a side effect of normal work.
Active ingestion. Skill executions create nodes automatically. Every time the user highlights text and runs a skill, the input, the output, and the entities extracted from both become nodes in the graph. The user's daily work becomes their knowledge base.
Deliberate ingestion. File drops (Kent Drop) and workspace file ingestion allow users to deliberately add documents, PDFs, and other files to their knowledge graph. This is for bulk import of existing work product.
The combination of passive, active, and deliberate ingestion means the knowledge graph grows continuously without requiring changes to the user's workflow. The professional works normally. Kent captures the knowledge.
4.2 Entity Resolution: Connecting the Dots
Raw ingestion is not enough. A knowledge graph with 10,000 disconnected nodes is a database, not intelligence. Kent's entity resolution pipeline connects nodes by identifying recurring entities -- people, projects, companies, concepts -- across different sources and creating edges between related nodes.
When a consultant mentions 'Henderson LLC' in an email, a meeting note, a proposal, and a deliverable, entity resolution recognizes that all four references point to the same entity and creates connections between them. When the consultant starts a new task involving Henderson LLC, all four sources surface automatically through the connected graph.
Entity resolution transforms a pile of facts into a web of knowledge. The web is what eliminates the blank page -- it enables the knowledge graph to surface relevant context from across all sources, not just the one the user happens to remember.
Conclusion
The blank page is an artifact of disconnected knowledge. It exists because the professional's relevant experience is scattered across a dozen applications, none of which talk to each other.
Kent connects them. Every email, every document, every conversation, every skill execution becomes a node in a unified knowledge graph. When a new task begins, the graph surfaces everything relevant -- automatically, instantly, and without requiring the professional to remember what they know or where they stored it.
The blank page is extinct. The pre-loaded page is here. And it gets richer with every day you use it.
References
- Microsoft. (2025). 'Work Trend Index 2025: The State of AI at Work.'
- McKinsey & Company. (2025). 'The State of AI in 2025: Application Fragmentation and Knowledge Worker Productivity.'
- Kent. (2026). 'Internal Usage Analytics: Context Builder Performance and Cold Start Reduction.'
Published by Kent Research, July 2026.