How I Closed a Six-Month Knowledge Gap in One Week: Using AI to Find What I Didn't Know I Missed

If you work in any fast-moving field, stepping away, even briefly, creates a real gap. Whether it’s an intense project, career transition, or life event, coming back often feels overwhelming. The same goes for anyone trying to break into something new: the sheer volume of what you don't know yet can be enough to stop you before you begin.

You don't just need to know what changed. You need to know what changed that matters. More importantly, you need to identify the things you don't yet know to look for.

Two career gaps, one clear winner for the easier transition back

Recently, I found myself in an unusual position: I effectively ran a controlled experiment on myself. I've been at the same company and in the same city for five years. I returned from maternity leave in 2022 and again in 2026. Same employer. Same colleagues. Similar responsibilities. But the experience of returning felt dramatically different. In 2022, catching up took months. I pieced together articles, meetings, conversations, and scattered notes. The process was fragmented, inefficient, and inevitably incomplete. This time, I rebuilt situational awareness and developed an action plan in about a week. The difference wasn't that less had changed. More had changed. The difference was that the tools for navigating change had matured alongside the field itself. And because I work at the intersection of AI strategy, data, and applied implementation, helping organizations build systems people actually use, I wanted to understand whether this approach generalized beyond my own experience.

It does. So now, I’m going to share what I did.

The Real Problem Isn't Information Overload, It's Blind Spots.

When I returned to work, I had a running list of articles and posts I'd bookmarked during leave. A list built on the eternal optimism that I would, at some future point, have both the time and the energy to read all of it. I also knew the list wasn’t comprehensive, it had an important structural flaw: it only contained things that had surfaced in my existing feeds and networks. It reflected my prior interests, not the full landscape.

The things I'd missed weren't the things I knew to look for. They were the things I didn't know I didn't know.

So the first thing I did was probably the most valuable: I asked AI to audit my blind spots.

I shared reading material I'd saved, along with context about my work: course materials, previous projects, consulting focus areas, and existing priorities. Then I asked a simple question:

Synthesize the major developments from the last 6 months across all sources. Organize by:

• new capabilities,

• new tools for builders,

• real-world applications,

• concerns and failures.

For each category, give me the 3–5 most important takeaways with specific examples.

“What important developments from the last 6 months aren't represented here?”

Why it worked: Reading about tools is not the same as understanding them. This prompt ensured I moved quickly from awareness to application.

The Most Important Shift I Observed

The development that surprised me most wasn't any single tool, it was how much agents had matured. AI has become significantly better at taking action, not just generating text. Where ChatGPT or Claude would previously produce a written summary of a workflow, AI can now execute many of those steps directly.

The practical implication: it's now genuinely feasible to build reusable pipelines for recurring work, with human checkpoints at the right moments. Each workflow you build compounds. You invest the time once and recover it every time you run it. The gap between "technically possible" and "practically usable" closed considerably while I was away, and I hadn't fully registered that until I looked at it all at once.

This Isn't Really About AI

I teach AI. I build with AI. I help organizations think strategically about AI adoption. Staying current is part of my job. And even I found returning disorienting. The lesson wasn't that experts don't fall behind. The lesson was that keeping up is increasingly about building systems rather than consuming more information.

The framework generalizes:

• Identify blind spots before consuming content

• Prioritize ruthlessly instead of reading linearly

• Use source-grounded tools when accuracy matters

• Reconcile old assumptions with current reality

• Move from awareness to application quickly

The challenge isn't simply the speed of change. It's building a system that helps you see what you've missed.

Based on all uploaded sources, create a prioritized reading list:

• MUST READ (30–60 min each): 3–5 sources essential for understanding the current landscape. Why each is critical.

• SKIM SELECTIVELY (15 min each): 5–7 sources for specific topics. What sections to focus on.

• REFERENCE AS NEEDED: remaining sources and when you'd use them.

The results were genuinely useful. One example surprised me. I'd completely missed how far AI evaluation frameworks (i.e., evals) had come. Evals are the frameworks used to measure how well an AI system performs, in other words, how do you know if your AI is actually performing reliably and safely? Not a glamorous topic, and it doesn't trend on social media. But it's increasingly critical for anyone building seriously with AI. I wouldn't have found it on my own. The AI surfaced it because I gave it enough context to know what mattered to me specifically, not just what was popular.

I added everything substantive and credible to a NotebookLM notebook and got to work.

Why Source Grounding Matters

I use Claude and ChatGPT constantly and could have used either for this. But for re-entry specifically, NotebookLM had one property I needed: source fidelity. I could trust it was drawing only on content I had uploaded. No hallucinated citations, no bleed from training data, no invented sources. Every answer linked directly back to the original material.

When you're trying to reconstruct a landscape you've been absent from, that kind of verifiability matters. I didn't need fast answers. I needed accurate ones I could check.

The Four Prompts That Did the Heavy Lifting

These are the actual prompts I used. You can adapt them for any domain: product strategy, investing, a technical field, whatever you're trying to catch up on.

1. Build a Structured Map of the Landscape

Why it worked: This created a navigable “map” of what actually changed, rather than a scattered list of updates.

2. Prioritize What Actually Deserves Attention

Why it worked: This one saved the most time. Instead of reading linearly, I had a triage system, a clear distinction between what to absorb fully and what to scan. If you only use one prompt from this list, use this one.

3. Figure out what's outdated, what’s new, and what’s evolved (e.g., reconcile old knowledge with new reality)

I last engaged deeply with this field 6 months ago. Based on all sources:

• OUTDATED: What should I remove or update? Tools or concepts now obsolete.

• NEW: What must I add? Capabilities that didn't exist, tools now essential.

• EVOLVED: What needs updating? Concepts that have shifted, tools with new features.

Why it worked: This is the prompt I'd most recommend to anyone returning from a gap. Rather than layering new information on top of old assumptions, it forces a direct comparison between your prior mental model and the current state. That's a meaningfully different kind of update.

4. Turn Knowledge into Action Quickly

Create an action plan for hands-on experience:

• Week 1 (2–3 hours): Which 2–3 tools should I try first? What should I build with each?

• Week 2 (2–3 hours): What needs deeper exploration?

• Week 3+: What should I be aware of but not master yet?

Interested in practical AI implementation, workflow design, or AI strategy? Book a consult today.