AI and the Challenge of Context Setting: Are We Helping Them More Than They Help Us?
Artificial Intelligence has transformed industries by offering automation, improved efficiency, and intelligent decision-making support. AI-powered assistants are positioned as tools that streamline workflows, providing structured insights, breaking down complex requests, and identifying potential impacts quickly.
However, the reality of working with AI assistants is not always smooth. While AI promises efficiency, using it effectively often requires significant human effort.
The assumption is that AI should eliminate inefficiencies and enhance productivity, but in many cases, businesses find themselves spending considerable time providing AI with the necessary context to function well. Instead of AI autonomously understanding a situation, users must input structured data every session, feeding AI the same critical details repeatedly. This raises an important question:
Is AI truly helping businesses become more efficient, or are humans spending more time managing AI’s limitations than actually benefiting from its capabilities?
NOTE: In this article we are going to use an example of working on enhancement request for a software application where we want AI to help us.
Understanding the Input AI Needs
To help AI effectively process enhancement requests for existing software application and generate actionable insights, product managers and business analyst must provide well-structured input that sets the right context.
Key Inputs for AI in Enhancement Requests
1. Contextual Information
AI needs a basic understanding of the application, its architecture, and the purpose behind the enhancement request to assess feasibility and alignment with existing functionality.
2. Detailed Enhancement Request
Clearly defining what the customer is asking, its urgency, dependencies, and technical feasibility helps AI determine the best approach for implementation.
3. Impact Analysis
Evaluating how the enhancement affects various aspects of the application, including performance, risk factors, and user experience, ensures a smooth integration.
4. Implementation Scope & Strategy
Defining the level of effort, possible approaches, testing requirements, and potential rollout strategies allows AI to recommend the most efficient path forward.
5. Integration & Deployment Plan
A well-structured release strategy, training needs, and post-launch monitoring ensure that the enhancement is successfully adopted and meets customer expectations.
This is just an example, but you can explore AI for assistance with any complex product, service, or application and experience its capabilities first-hand. However, you will often find that providing detailed contextual information is essential for AI to deliver meaningful and accurate support.
This challenge becomes even more pronounced in dynamic team environments where people join and leave projects, roles are handed over to new team members, and outsourcing or consulting partnerships introduce fresh perspectives. Each individual interacting with AI may phrase requests differently, provide varying levels of detail, and interpret AI-generated insights in their own way.
In large teams working on complex products, the inconsistency in how AI is engaged can lead to fragmented understanding, requiring structured processes to maintain continuity and ensure AI-driven responses align with the broader team’s knowledge and objectives.
AI as an Instant Processor—But Not a Long-Term Thinker
One of AI’s strongest attributes is its ability to process structured information instantly. Given a direct query, an AI chatbot can:
- Break down enhancement requests into smaller components for easier implementation.
- Analyse impacted features and dependencies in complex applications.
- Provide structured responses quickly, eliminating long brainstorming sessions.
- Offer multiple perspectives to help teams make informed decisions.
At first glance, this seems revolutionary—a tool that enhances efficiency, precision, and speed.
However, AI assistants lack something fundamental: the ability to retain long-term knowledge. Unlike human employees who develop expertise over time and refine processes with each experience, AI begins every session from scratch.
This means every time a user engages with an AI assistant, they need to reintroduce details about their product, previous enhancement requests, dependencies, and relevant risks because AI has no recollection of past interactions. Without session-to-session memory, AI struggles to function as a truly effective collaborator.
The Effort Required for Context Setting
Humans working in an organization naturally remember past decisions, product iterations, and customer feedback trends. AI, on the other hand, does not recall prior discussions, making long-term collaboration challenging. This forces businesses to dedicate time to continuously setting context, an effort that contradicts the efficiency AI was supposed to provide.
Before AI can meaningfully contribute to enhancement requests or product feature analysis, users must manually track and provide:
- Product architecture details – AI must be informed about how features interact.
- Recent product changes – AI won’t remember past updates unless explicitly outlined.
- Customer requests and feedback – AI can’t recognize patterns unless they are provided every session.
- Potential risks and dependencies – Without past discussions, AI can’t anticipate challenges.
This becomes an unexpected time-consuming task—one that businesses must manage to keep AI relevant.
The Boundaries of AI in Contextual Analysis
AI has transformed the way businesses process data and make informed decisions. However, despite its impressive capabilities, AI operates within strict limitations that restrict its ability to analyze and interpret non-textual inputs. Understanding these constraints helps businesses set realistic expectations when leveraging AI for contextual analysis.
Key Limitations of AI in Contextual Understanding
1. Inability to take URLs as input
No, AI chatbots cannot directly browse or analyse multiple URLs as input for setting context. They do not independently visit webpages or extract content from them. Instead, they rely on structured textual input provided by users, such as summaries or key details from those web pages.
2. Inability to Analyse YouTube Videos Directly
AI cannot watch or comprehend product demo videos by simply receiving a YouTube channel URL. It requires structured descriptions, transcripts, or summaries to understand the content.
3. Restricted Access to SharePoint Documentation
AI cannot automatically retrieve and process documents stored in SharePoint sites that requires authentication. Accessing such materials requires manual input or summaries from users.
4. Lack of Visual Interpretation for Architecture Diagrams
AI is unable to analyse Visio diagrams or PowerPoint slides containing product architecture and data structure details. Text-based explanations are needed to convey the relevant information.
5. Inability to Process 3D CAD Models
AI does not interpret CAD designs, making it unable to assess product design through 3D models. Any structural insights must be provided in a textual format.
6. Dependence on Structured Text Inputs
AI primarily relies on structured text data, meaning it struggles with ambiguity, visual elements, and experiential knowledge that humans naturally process.
Are Humans Actually Helping AI More Than AI Helps Humans?
AI was supposed to eliminate inefficiencies, but the reality of reintroducing context manually creates an ironic situation:
Are humans doing extra work just to make AI function properly?
A major expectation of AI is that it should do the hard work for humans — extracting insights, predicting impacts, and making logical recommendations. But in practice, AI users must become the AI’s helper, feeding it information in a way that feels more like training a new employee than working with an intelligent system.
The Frustration of Context Setting: Time vs. Productivity
If an AI assistant retained memory, a product manager could log enhancement requests once and receive continuous recommendations across sessions. However, because AI forgets everything, that same manager must re-enter every enhancement request during each conversation.
Over time, this inefficiency compounds, leading to lost productivity, frustration, and repetitive administrative work — precisely what AI was meant to eliminate.
Without persistent memory, AI cannot function as a fully integrated team member. It remains a single-session assistant, useful for short-term problem-solving but not long-term strategic planning.
Strategies for Making AI Useful Despite Its Memory Limitations
Businesses can still make AI work for them—but only if they actively adopt structured workflows to minimize the burden of reintroducing information.
Providing Contextual Information in JSON, Markdown, or JSON Schema Format
Structured formats like JSON, Markdown, or JSON Schema allow AI to quickly interpret requirements without ambiguity. These formats help standardize contextual data, making it easier for AI to parse and generate accurate insights. Products like 🏵 OpenRose can streamline this process, ensuring AI receives well-structured inputs that improve efficiency and precision in analysis.
Maintaining a Centralized Knowledge Repository
Instead of repeating everything manually, store ongoing project details in a shared document. This becomes the go-to reference point that can be quickly accessed whenever AI assistance is needed.
Using Structured Summaries in Every Session
Rather than explaining product details from scratch, start each AI conversation with a concise update. Instead of full historical context, summarize only what has changed since the last enhancement request.
Uploading Visual Aids for Faster Processing
Instead of writing long explanations, upload screenshots, architecture diagrams, or workflow visuals. AI can process images, extracting meaningful insights without requiring extensive text-based descriptions.
Leveraging AI for Structured Analysis, Not Long-Term Planning
AI excels at structured insights but struggles with continuity. Businesses should use AI to:
- Break down enhancement requests into manageable steps.
- Analyse feature dependencies and risks for short-term fixes.
- Summarize structured discussions into action points—rather than expecting AI to track project evolution.
Using AI for Brainstorming Instead of Execution
Since AI lacks long-term memory, use it for idea generation rather than ongoing planning. If AI can’t retain historical data, it can still be valuable for generating fresh perspectives, exploring alternatives, and outlining new solutions based on present-day input.
The Future of AI: Persistent Memory and Adaptive Learning
AI will only truly transform businesses once it can retain session-to-session knowledge, eliminating the need for repetitive context setting.
Persistent AI memory would allow businesses to:
- Track enhancement requests over time without manual input
- Build institutional knowledge inside AI assistants, reducing human effort
- Enhance AI-human collaboration, where AI proactively recognizes dependencies without requiring explicit input
AI shouldn’t create more work for humans —it should seamlessly integrate into workflows without requiring manual intervention.
Final Thoughts
For now, AI is a useful tool, but one with significant limitations. Businesses need to weigh its benefits against the hidden cost of continuous manual input. AI is not yet an autonomous team member—it still relies on human interaction to remain effective.
Organizations must adopt structured workflows that help AI remain useful without requiring excessive manual input. AI can certainly improve efficiency when used correctly, but businesses should be aware of the hidden costs associated with context setting.
For AI assistants to truly integrate into workflows, they must evolve into systems that autonomously learn, adapt, and retain institutional knowledge. Until then, AI will remain a valuable but limited tool—capable of offering structured insights in the short term but requiring extensive human effort to maintain continuity.
THIS BLOG POST TITLE COULD HAVE BEEN
Comments
Post a Comment