Google’s AI-powered study tool, NotebookLM, has become a game-changer for me when I need to learn new skills. The combination of guardrails, the reference material you feed it, and the Gemini-powered chatbot makes it a powerful ally in comprehension. But what about skills that I already know?
Can the AI model help with those, like building a PC, something I’ve done dozens of times myself and hundreds of times for friends? Can it synthesize data to produce a similar recommendation list to the ones I’d pick personally, or would it fall short of the stated aim? Well, after prodding at NotebookLM for an afternoon, I can say that it’s almost as good. And with a little more prompting know-how, I could probably get it to make ae…
Google’s AI-powered study tool, NotebookLM, has become a game-changer for me when I need to learn new skills. The combination of guardrails, the reference material you feed it, and the Gemini-powered chatbot makes it a powerful ally in comprehension. But what about skills that I already know?
Can the AI model help with those, like building a PC, something I’ve done dozens of times myself and hundreds of times for friends? Can it synthesize data to produce a similar recommendation list to the ones I’d pick personally, or would it fall short of the stated aim? Well, after prodding at NotebookLM for an afternoon, I can say that it’s almost as good. And with a little more prompting know-how, I could probably get it to make aesthetic choices based on a list of personal preferences.
The AI research assistant inside NotebookLM is my best buddy
It’s like having a study buddy for anything you could want
NotebookLM is part knowledge repository, part eager intern, and part wise old sage, with features that might surprise you. It’s not designed to do the work for you, but it will help you make sense of what you’re studying, find additional resources, and organize them into an easily digestible format.
One day, I might be feeding it spreadsheets and asking for analysis, or product specifications to cross-reference against benchmarking results. Or help me debug some code, reformat YAML, or figure out the best workflow for setting up complex Docker stacks. The point is that it works with me, not for me, and hasn’t (so far) hallucinated any results that I know to be wrong.
Yes, including if I want to build a new PC
If you’re thinking about building a new PC, you might read sites like XDA, trawl YouTube for techtubers’ builds, or use sites like PCPartPicker to narrow down your choices while staying within your budget. And, well... so does NotebookLM, if you feed those source sites into a notebook.
The tool will then turn those into mind maps, user flows, and a repository of information that can be queried using natural-language queries. If what you ask for isn’t there, it’ll tell you it doesn’t exist in the sources, which is a good prompt to go find more sources to add, so the end results are better.
Planning a PC build with NotebookLM was fun
And I learned a few things
I’m a very visual learner, but I don’t always know how to make my notes in formats that help, especially when I get stuck in analysis paralysis. But NotebookLM doesn’t suffer from that, and turned the PC building resources I gave it into a comprehensive mind map, breaking it down into budget, components, cooling and airflow, and even performance tweaks and troubleshooting steps. These could all be followed through a series of logical leaps to get into the weeds of how to pick a suitable collection of PC parts for any type of build, and it made so much sense.
Things that I know but can’t always articulate, such as why you’d allocate for specific components like your GPU before other parts, or the merits of prebuilt PCs compared to building your own were all laid out in plain language to read through. If I didn’t have a grounding in PC building, I would still have found it easy to follow, leading to a usable parts list to source.
It does more than just suggest parts
Because it has internet access, NotebookLM can tell you current pricing, or historical trend lines. But it’s even more powerful when you ask questions about the wider ecosystem. I wondered which current trends and events are affecting prices and performance, and the AI gave me a comprehensive rundown of the current state of things. These are insights that I would have needed hours of searching, collating, and spreadsheet analysis to reach, all gleaned from the few dozen websites I fed into the sources.
It’s also a little unhinged
I’m not sure where it plucked this little morsel of an analogy from, but it needed a little longer in the oven. I’m not even sure the kitchen is the right comparison when it comes to PC builds, but NotebookLM had a go anyway.
Analogy: If PC building tiers were like cooking supplies, the sub-1,000tierislikebuyingstarterpotsandpans—functionalformostbasicmeals.The∗∗999 – $1,399 Entry-Level tier** is where you invest in a quality, modern induction stovetop (AM5/DDR5 platform) and reliable mid-grade chef’s knives (RTX 5060 Ti) that will give you excellent results right away and can easily accommodate a fancy new blender or specialized cookware (future CPU/GPU upgrades) without needing to replace the whole kitchen base.
It was fun, though, and a refreshing change from the fawning demeanor of ChatGPT and its ilk. I feel tools like NotebookLM are where the magic is in AI, where the model is there but relying on information you feed it to put together, rather than making up its own sources.
It planned a fairly decent PC
I asked for help building a gaming PC that could do 1440p at high refresh rates, with a rough budget of $1,500. And that’s what I got —no missing components, and with most of the brands and parts, I might well have picked them myself from the PCPartPicker lists based on pricing, performance, and availability.
(Reproduced verbatim from NotebookLM)
| Component Type | Model | Price (at Pub Time in USD) | Key Function / Goal | ||
| CPU | AMD Ryzen 7 7800X3D | $359 | 8 cores/16 threads with massive L3 cache for elite gaming | ||
| Motherboard | MSI MAG B850 Tomahawk MAX Wifi | $210 | Best budget AM5 board, supporting Wi-Fi 7, future CPUs via BIOS flashback | ||
| GPU | RTX 5060 Ti (16GB) | $429 | Provides brilliant 1080p and strong 1440p gaming; the 16GB model is recommended over the 8GB model for VRAM longevity | ||
| RAM | Crucial Pro 32GB (2x16GB) DDR5-6000 | $145 | 32GB is the sweet spot for high-performance gaming and streaming.DDR5-6000 is the beginning of AMD’s ‘sweetspot’ for memory speed | ||
| Storage | WD Black SN850X (2TB) | $158 | One of the best PCIe 4.0 SSDs available, providing sufficient 2TB capacity for several AAA games | ||
| Case | Lian Li Lancool 216RX | $100 | Excellent airflow design with dual RGB 160 mm front and single 140 mm exhaust fans | ||
| PSU | ASRock Steel Legend SL-750G 750W 80+ Gold | $95 | 80 Plus Gold certified, supporting the ATX 3.1/PCIe 5.1 standards with a native 12V-2x6 cable for the GPU | ||
| Cooler | ID-Cooling FrostFlow 280mm | $56 | An inexpensive AIO cooler to manage the CPU temperature | ||
| TOTAL | N/A | $1,552 | N/A |
It even took a little license to overstep the budget by $52 and did a good job of selecting parts that would work well together, fit inside the chosen chassis, and pose no blockers.
NotebookLM is only as good as its sources (but I did get a PC planned out)
I’m going to incorporate NotebookLM into more of my buying decisions, especially for household appliances and other areas where I’m less up to speed. It’s not just that I can ask for initial analysis of things, but that I can ask qualifying questions or further angles of inquiry to really drill down on the specifics of a choice, and know that it’s based on sources I trust.