Stop Re-Explaining Context to AI - How Projects Grow Themselves

Every AI conversation starts the same way: explaining context you've already explained dozens of times before. "I'm working on a product that does X, my team structure is Y, and I need help with Z." By the time you finish setting the stage, you've burned half your thinking time on background instead of solving the actual problem.

I used to treat AI projects like static file folders - dump some context in there and reference it occasionally. Then I discovered something counterintuitive: the best AI projects don't just store information, they grow themselves.

Here's how I transformed my AI workflow from repetitive context-setting to a self-evolving knowledge system that saves me hours every week.

The Two Types of Project Content

After building dozens of AI projects across multiple clients, I've found there are exactly two types of content that matter:

Reference Content: The contextual foundation that informs every conversation. This includes your business model, team structure, technical architecture, and strategic goals. Think of it as the background knowledge that makes every AI response more relevant.

Hotkeys: Automated workflows that execute complex tasks with simple triggers. These are your recurring processes - checking yesterday's Jira tickets, evaluating sprint progress, analyzing Slack conversations for issues - all packaged into reusable macros.

Most people only use reference content. That means they're missing out on the biggest productivity gains. McKinsey research indicates that 60% of employees could save 30% of their time with workflow automation. That's exactly what sophisticated hotkeys provide.

How Projects Actually Grow Themselves

Here's the workflow that changed everything for me:

Start with a seed: Load your initial context - product idea, client background, or project scope.

Build through conversations: Every significant AI discussion should end with an artifact or canvas. Don't just chat and close the tab.

Feed results back: Take that artifact and add it to your project knowledge. Now your next conversation has both the original context and the new insights.

Repeat and compound: Each iteration builds on previous work. Your product idea grows marketing strategy, then customer profiles, then market analysis. All of this becomes accessible in future conversations.

The key insight: each conversation becomes the foundation for better conversations. Instead of starting from scratch, you're building on increasingly sophisticated context. This approach aligns with findings that workers using generative AI tools can boost their performance by up to 40% compared to those without. But this only works when the AI has rich, evolving context to work with.

The Quality Control Secret

Here's where most people fail: they dump raw AI output into their projects and wonder why everything becomes generic nonsense.

The rule: Only refined artifacts go into project knowledge.

This means actively editing and improving AI-generated content throughout your conversation. Use the artifact or canvas features to iterate: read it, critique it, update it, refine it. The final version that goes into your project should be something you'd actually reference.

Bad: Copy-pasting first-draft AI responses into your project. Good: Spending time refining artifacts until they represent your actual thinking.

Raw AI output dilutes your ideas. Refined artifacts amplify them. This distinction between productive work and busy work mirrors what we see in traditional project management. Understanding the difference between doing good work and doing the right work becomes crucial when building AI systems that actually serve your goals.

The hallucination trap: When you don't review artifacts carefully, hallucinated details slip into your project knowledge. Then future conversations treat those false details as fact, compounding the problem. I've seen AI confidently reference "features" that don't exist because an earlier unreviewed artifact made them up.

Regular pruning matters: As projects evolve, you'll build artifacts that overlap or contradict each other. Schedule monthly consolidation sessions to merge related content, eliminate outdated information, and reduce total context size. Your most recent conversations should build on your best thinking, not compete with obsolete artifacts.

Building Your Hotkey Library

Hotkeys are where the real time savings happen. Instead of typing "check yesterday's Jira tickets for critical issues using this specific JQL query and evaluate them based on these criteria," I just type "check yesterday's tickets."

That's it. Three words instead of a paragraph.

The productivity impact is substantial. Studies show that 80% of knowledge management system users save at least one hour per week by reducing time spent searching for information—exactly what well-designed hotkeys accomplish. The average knowledge worker currently spends 19% of their workweek searching for information, but sophisticated hotkeys can eliminate most of this information hunting.

The hotkey contains:

• The specific URL or query to run
• The evaluation criteria I care about
• The format I want results in
• Any MCP server commands needed

Examples from my current rotation:

• Daily Jira ticket triage with criticality assessment (using Atlassian MCP server)
• Weekly sprint status evaluation across multiple projects (similar to analyzing burndown charts for sprint insights, but automated)
• Slack message analysis for escalation patterns (building on the communication insights from analyzing Slack metrics)
• Architecture decision tracking across client conversations

As a project manager across multiple clients, these hotkeys save me about an hour every morning. I run about ten of them in sequence. Within ten minutes, I have complete status across all organizations.

Instead of manually reviewing hundreds of Jira tickets, the AI identifies the 3-4 that actually need my attention. Instead of scrolling through Slack channels, I get analysis of key conversations and potential escalation patterns.

The automation benefits mirror broader workplace trends. Research indicates that workflow automation can reduce repetitive tasks by 60-95%, leading to time savings of up to 77% on routine activities. This isn't theoretical. It's measurable productivity improvement that compounds daily.

AI tools excel at consuming and synthesizing information. I'm not using them to produce content like Jira comments - for intelligent filtering and prioritization, they're incredibly effective. This filtering approach aligns with effective priority management for development teams, where the goal is identifying what truly needs attention rather than trying to address everything at once. If you ask AI to find "important" tickets, it will struggle, but if you create a reference document defining what important means, it's fairly good at finding those types of tickets.

The Tool-Agnostic Strategy

The biggest mistake is storing everything inside Claude or ChatGPT projects. You're one platform change away from losing all your work.

I store all reference content and hotkeys in Obsidian, then load them into both Claude and ChatGPT projects. This gives me:

• Platform independence
• Easy backup and version control
• Ability to use the same context across different AI tools
• Migration flexibility when better tools emerge

Your project knowledge should outlive any single AI platform.

Team collaboration bonus: Store your project knowledge in shared folders (Google Drive, shared Obsidian vaults, etc.) so team members can load the same context into their preferred AI tools. Everyone works from identical background information, even using different platforms.

This approach gets complex when content updates frequently, but the alignment benefits justify the coordination overhead. Organizations that provide AI-based tools and training report over 10% increases in annual revenue compared to those that don't. This happens partly because coordinated AI usage amplifies individual productivity gains across teams.

The Unexpected Mobile Advantage

Once you've built rich project context, AI mobile apps become incredibly powerful. It's nearly impossible to type complex context on a phone, but with everything pre-loaded, you can have sophisticated conversations anywhere.

This aligns with broader voice AI adoption trends. The voice recognition market is expected to reach $53.94 billion by 2030, with 71% of people already using voice assistants for research and productivity tasks. The workplace applications are particularly compelling. Voice AI is becoming essential for productivity, allowing employees to work smarter through automating routine tasks and decision-making support.

I regularly use voice input to Claude on my phone for quick project updates, knowing all the context is already there. The AI understands my clients, processes, and preferences without needing explanation. With 41% of US adults using voice search daily, this mobile-first AI interaction is becoming standard practice for knowledge workers.

Start Growing Your Next Project

Pick one recurring work challenge you have. Instead of solving it once and moving on, build it into a growing AI project:

1. Create the project with your basic context
2. Define one hotkey for a task you do weekly
3. Have a conversation that ends with a refined artifact
4. Add that artifact to project knowledge
5. Test the hotkey with a simple trigger phrase

Within a week, you'll have a project that's more useful than anything you've built before. And it will keep getting better every time you use it.

The compound benefits are measurable. Consider that employees estimate automating tasks could save them 240 hours per year, while company leaders believe the savings could reach 360 hours annually. With 75% of workers now using AI in the workplace, those who build evolving AI systems rather than starting fresh each time will capture disproportionate productivity gains.

Multi-client scaling: If you work across multiple organizations, create one project per client. Each contains that client's architecture, team structure, tools, and processes. Every conversation happens within the appropriate project context - I rarely have "vanilla" chats anymore.

This approach scales efficiently whether you're managing multiple projects within one company or consulting across multiple clients. The context-switching research is particularly relevant here: it takes an average of 25 minutes to fully return to productive work following an interruption. Pre-loaded project contexts eliminate this refocusing penalty entirely. This connects directly to foundational time management principles. By removing the cognitive overhead of context-switching, you can focus on higher-level strategic thinking rather than constantly reorienting yourself.

Your AI tools should learn and evolve with your work, not force you to start over every conversation.

Stop re-explaining context. Start building systems that remember everything and do the work for you.

---

Connsulting Blog RSS

Related Content