Editor's note: This article originally appeared on LinkedIn
One of our core values at LinkedIn over the past 20+ years is that we put our members first. This value has been central to how we've built our platform, and it guides us when making every business and technical decision. The type of conversations we foster between our members, the kind of products we build for our customers, and the engagement we seek to drive on our platform all stem from a desire to do what matters most in the world of work.
This principle has guided us to create a massively scalable platform to connect our members to jobs, skills, knowledge, and buyers. LinkedIn is the world’s largest economic opportunity platform - and this focus will continue to be our north star in the new era of AI, and AI Agents in particular.
So how do we do this?
We’ve used AI for over a decade at LinkedIn to help our members achieve their goals. Underpinning this intelligence is our complex technology stack, thousands of engineers globally, and the power of the Economic Graph - a digital representation of the global economy with data from over one billion members, 68 million companies and 41 thousand skills.
The economic graph consists of information provided by our members and from public sources. This is why the information you share on LinkedIn is so important. That your profile is complete and verified matters. It’s why we want our understanding of the world of work to be as deep as possible. And it’s why we take the security and privacy of the data members entrust to us as such a serious responsibility.
Combining all this enables us to create a hyper-personalized experience and understand our members. This gives us the knowledge needed to improve our matching technologies, which allows us to show a relevant job being presented to a job seeker or help them better search results for a member seeking knowledge on a complex question.
With the rapid advancements in AI, we can do even more of this. And the launch of ChatGPT changed the game. Overnight, member and customer expectations grew to include AI as a prominent part of their daily experiences. We responded swiftly, integrating Generative AI visibly and usefully into our products. For example, AI-powered writing assistants that help craft your profile and InMail messages, and there’s lots more we will do like bringing some of this power to how you find what you need on LinkedIn.
This idea of personalization in professional work is a concept I’ve thought about for a long time - starting from my very first professional experiences. In my first college internship, my mentor introduced me to Emacs - an incredibly powerful text editor. I built a custom .emacs file - a way to personalize Emacs and introduce my specific key bindings, my auto-correct, my preferred way of working. It has my slang and my lingo built into it. I customized it so that whenever I type my name, it knows there is no “w” in Mohak! My needs for it have expanded over time, I’ve added to it and continued to add to it. I used Emacs as long as I could throughout my career, and carried that .emacs file from job to job, using it to find a flavor of consistency between very different corporate environments and tools.
I eventually outgrew Emacs, and it no longer fit into my regular flow of work. And the loss of my .emacs file (I still keep it around for sentimental reasons - it's fun to fire up Emacs now and again to view some code and feel that familiar environment again) is something I regularly think about - how I’ve lost the ability to personalize my work tools the way I did before. But technology is evolving again in a direction that can return truly personalized work to all of us.
As we enter the agentic era, our guiding principle is to continue to create economic opportunity and put our members first through agents personalized to you - that work where you work, how you work, and enhance opportunity for you. While this new technological frontier is like nothing we have experienced before, what isn’t new is that LinkedIn will remain a trusted partner our members and customers can rely on to help them in their professional lives.
This means we are building agents that should be efficient task-doers supervised by you, picking up the tasks you don’t want to do so that you can focus on what you do best. Whether that’s helping a recruiter hire for the right role, a marketer be more successful or a learner building the right skill. We will put you, the professional, in the driver’s seat giving the agent directions for the tasks you need completed and providing oversight on actions so that it can learn your preferences and make interactions more personalized to how you work. This will give you more time and space to focus on what you love about your job, increasing your access to opportunity. We see this as giving our members and customers agency in the agentic era.
In this new era of professional life, our members will interact with a variety of agents daily, similar to how we currently use multiple apps in our everyday life. While I use over 20 apps daily now; in the coming years, these will be agents, all vying for my attention. For many professionals, this will mean massive cognitive overload.
I imagine a future where our members and customers have this level of control over the agentic team working for them. This takes me back to that .emacs file - rich customization, personalization, and comfort. In this future everyone will have their own digital chief of staff, that understands their preferences, injects their personality, and helps manage their team of agents by involving them to supervise when necessary.
We’re already building technology to support this next era. We recently developed an innovative new piece of infrastructure that I’m really excited about - experi ential memory - that allows agents to become more personalized and adapt to the unique skills of each professional, learning from their feedback and adjusting how they work. This capability is key to how we’re going to make an agent your personal digital chief of staff.
We also built a new orchestration layer that uses the reasoning abilities of LLMs to organize and act through interactions with recruiters and support from tools that enable things like search and messaging. This layer helps us take a real-world approach to tasks - iterative, asynchronous and collaborative and will allow this new technology to be a tool for humans. Our real identity platform and verification systems also mean that we have credibility in this space. Our long history of connecting members and customers with technology, and experience in this space is also helping us build this new technology.
Aarathi Vidyasagar captured this well when we announced our first agent, LinkedIn Hiring Assistant, which is designed to take on a recruiter’s most repetitive tasks so they can spend more time on their most impactful work.
And we’re just getting started. Our engineers are passionate about this technology and exploring the various ways they can help our members and customers. We are trying and testing to manifest those ambitions and form more agents that we will bring to our platform in the coming months.
Like anything we build at LinkedIn, we will create agentic experiences that are personalized to you, with a focus on your goals. We believe that this new era, approached in this way, will enable far greater professional agency for every member of the global workforce.