Google has made this more useful
Google’s new notebook integration between Gemini and NotebookLM is a real improvement.
Not because it changes everything. Not because it suddenly turns Google into the answer to long-term knowledge work. But because it makes the workflow more coherent than it was before. Chat, sources, notes and outputs are starting to sit closer together, which is a good deal more useful than the old routine of uploading a few files into a model and hoping it behaves itself.
For research-heavy work, that matters. If you spend your time reading, comparing sources, synthesising information, or trying to turn messy material into something usable, Google has made this easier.
That is worth acknowledging.
Why people are overstating what this means
The usual launch-week habit, of course, is to take a genuine product improvement and turn it into a grand declaration that one tool has now killed another.
It has not.
Google has improved its AI research environment. It has not built a replacement for a long-term thinking system, and it certainly has not removed the need for judgement. That is the broader problem with how people talk about software in search and marketing: they keep confusing tools with strategy, which is usually where the real work begins. That is true here as well, and it is the same reason many businesses end up needing strategic SEO consulting rather than another round of enthusiastic tooling advice.
Google’s notebooks are built for projects
A Google notebook is a container for a defined piece of work.
You gather a set of sources, ask questions, generate summaries, and move toward an output. That output might be a brief, a study guide, a presentation asset, or a digest of a topic. The point is that the system is built around focused synthesis.
That is where it looks strongest.
For plenty of users, that will be enough. More than enough, probably. Most people do not want to build an elaborate note architecture. They want to collect material, interrogate it sensibly, and get something useful back without too much friction.
Google is getting better at serving that need.
Obsidian is built for accumulated thinking
Obsidian does a different job.
It is not mainly about processing one source set at a time. It is about keeping ideas alive long enough for them to become more useful. Notes from a year ago can suddenly matter again. A passing thought can connect to a client problem months later. Something half-formed can turn into a framework, a decision, or a piece of work because the system lets ideas persist and collide over time.
That is the real value.
Not the “second brain” language. Not the plugin hobbyism. Just continuity. A place where thinking can accumulate rather than disappear once the immediate project is finished.
The real limitation in Google’s model
This is why the current excitement needs a bit of restraint.
Even with the new Gemini integration, Google’s notebook model is still organised around the notebook as the unit of work. That makes it useful for source-based research. It does not make it a durable personal knowledge layer.
Those are different things.
A project workspace helps you complete a task. A knowledge system helps you build a body of thought that remains useful beyond that task. Google is getting better at the first. Obsidian is still better suited to the second.
That distinction matters beyond note-taking software, because it points to a larger issue in AI-driven search and discovery: convenience is not the same as durable visibility, and synthesis is not the same as understanding. The businesses that will hold up best in this environment are the ones that can be clearly interpreted, retrieved and cited by these systems, which is precisely the kind of question a AI search audit is meant to answer.
The sensible conclusion
So this is not really a story about Google replacing Obsidian.
It is a story about Google building a more credible AI workbench for source-based research, while Obsidian remains better suited to long-term knowledge work.
Google deserves credit for making this better. It is more practical than it was, and more useful for people who want help turning material into outputs quickly.
But improved is not the same as equivalent.
At the moment, Google looks stronger on convenience and guided synthesis. Obsidian still looks stronger on continuity, control and accumulated thinking over time.
That is the more useful reading.
Less dramatic than the launch chatter, perhaps. Usually more accurate.