We used GPT-5.6 Sol’s browser use functionality, accessed via the ChatGPT/Codex desktop app. This functionality permits AI to basically take over the browser and stroll via the steps of laptop use as a human would, navigating web sites, clicking on hyperlinks and extra.
This is step-by-step how the AI did it:
- Opened the LinkedIn put up in a signed-in browser session. No handbook navigation required
- Switched from the default “most relevant” view to the view that confirmed the complete remark thread
- Started on the first remark and labored systematically via the web page in order
- Loaded further feedback as they appeared, scrolling and triggering LinkedIn’s “load more” conduct routinely
- Expanded replies inside remark threads with out being prompted to achieve this
- Opened feedback hidden behind LinkedIn’s “See More” button — that means longer feedback which might be truncated in the default view have been learn in full
- Separated Paul’s personal replies from the viewers’s feedback to keep away from skewing the evaluation
- Distinguished particular person commenters from the quantity of feedback they posted to keep away from duplicating one individual’s opinion.
- Grouped responses into “for” and “against” positions on the NYC faculty AI ban
- Surfaced recurring arguments that appeared throughout a number of feedback
- Pulled particular examples that illustrated the vary of views
Throughout the complete course of, a human watched because the AI labored. It’s not suggested to stroll away whereas an AI agent is working inside a dwell, signed-in account.
What the Results Showed
The closing evaluation landed at roughly 50/50 or almost an identical constructive and destructive sentiment towards Roetzer’s put up and the broader place it represented on AI in faculties.
Why This Matters for Marketers
Social listening is just not a brand new idea. What is new is the flexibility to do it rapidly, effectively and precisely with AI.
This makes the barrier between “I want to know what my audience is saying” and “here is what your audience is saying” considerably decrease.
Consider what this similar strategy might do throughout different advertising and marketing workflows:
- Analyze viewers sentiment on a product launch put up with out ready for a proper survey
- Track recurring objections in feedback on a competitor’s content material to inform messaging technique
- Monitor neighborhood responses to a marketing campaign earlier than deciding whether or not to amplify or modify
- Understand what arguments hold surfacing round an business debate your model has a stake in
- Separate the sign from the noise on any high-volume remark thread the place the loudest voices do not essentially signify the bulk
The underlying precept is straightforward: wherever a marketer would in any other case copy-paste content material from a browser to analyze it, AI browser use can take over that course of.
An Important Caveat
This sort of functionality comes with accountability connected.
Browser use — AI that operates inside a dwell, signed-in session — requires care. It ought to be run in a managed atmosphere, actively monitored all through, and scoped narrowly to the particular process at hand. The applicable posture proper now could be supervised and deliberate, not automated and unattended.
We ran our experiment as a really restricted take a look at. It’s necessary to word that sure websites, together with LinkedIn, have phrases of service that would ban your account in the event you apply this automation broadly or at scale.
We share this instance, in its restricted use, to assist folks perceive the numerous capabilities of AI browser use and its means to change what you are able to do in advertising and marketing.
Raising the Ceiling of What’s Possible
What’s doable with AI is rising quick. GPT-5.6 Sol dealt with this process with a degree of accuracy and autonomy that may have been shocking even a couple of months in the past. GPT-6 Astra, OpenAI’s latest launch, which was not but accessible on the time of this take a look at, is particularly positioned as a significant leap ahead in laptop use pace, accuracy, and reliability.
Marketers who begin experimenting fastidiously with this now in managed situations shall be forward of the curve when these capabilities grow to be the usual.
This put up attracts on the AI Use Case Spotlight phase of Episode 237 of The Artificial Intelligence Show, hosted by Paul Roetzer and Mike Kaput. To pay attention to the complete episode, go to: https://podcast.smarterx.ai/shownotes/237
