Whenever I talk with engineers about AI in recruiting, I usually get the same question.
“So…how are recruiters using AI to source candidates and review resumes?”
Fair question.
Sourcing and resume review get most of the attention because they’re obvious pain points. But after spending time talking with talent leaders who are actually building with AI every day, I’ve realized the most interesting AI workflows are helping remove the friction that keeps recruiters from doing their best work and allows them to actually spend more time recruiting and with their candidates.
One conversation that really reinforced this for me was with Aston, a TA leader who has been building internal AI workflows for his recruiting team. While every build he shared was different, they all had one thing in common: helping recruiters spend less time on administrative work and more time creating value.
It Starts Before The Search Ever Begins
One of the workflows Aston built acts as a research analyst before a client kickoff.
Instead of walking into an intake meeting with only a job description, the workflow gathers market intelligence, competitor hiring trends, talent supply and demand, and other publicly available insights to help shape the conversation.
That stands out because so many candidate frustrations start with unclear expectations. Recruiters can’t accurately communicate a role if the hiring team isn’t aligned on what they’re actually looking for.
If you improve the kickoff, you improve everything that follows.
AI Is Helping Create Better Hiring Alignment
The next workflow takes intake notes, the job description, and additional context to build an ideal candidate profile, generate structured screening questions, and create interview scorecards.
Again, this isn’t replacing recruiter judgment but creating consistency.
Instead of every recruiter interpreting a job differently, everyone starts from the same understanding of what success looks like.
That means candidates receive a more consistent experience, interviewers evaluate against the same criteria, and hiring teams spend less time debating what they actually want halfway through the search. This only helps the overall candidate experience.
Finally…Someone Explained MCPs In A Way That Made Sense
I’ll admit it.
I’ve heard people talk about MCPs for months, but I’d never actually built one.
So I asked Aston to explain it in simple terms.
His analogy:
“If the AI is the chef, your data is the pantry, and the MCP is the door between them that stays open while you’re cooking.”
That clicked.
I’ve spent plenty of time copying and pasting notes, job descriptions, transcripts, and documents into AI tools just to give them enough context.
With an MCP, that context stays connected directly to the source instead.
Less copying. Less manual work. Better quality outputs because the AI is working from live information instead of whatever I happened to paste into a prompt.
I’ve Completely Changed My Mind About AI Notetakers
I used to think AI notetakers were unnecessary.
My mindset was simple:
“I pay attention.”
“I take good notes.”
“I have a good memory.”
Then I started using one.
What surprised me wasn’t just the quality of the summaries but everything I missed.
Technical conversations move quickly, and no matter how focused you are, it’s impossible to capture every detail when you aren’t an engineer.
Having AI capture the conversation lets me focus on the person sitting across from me.
That was one of Aston’s biggest takeaways as well. His team uses meeting data not only to improve conversations but also to uncover patterns across interviews, identify missed opportunities, and surface market insights they can take back to clients. Once you stop thinking of meeting notes as documentation and start thinking of them as searchable data, the possibilities expand pretty quickly.
I think we are moving past folks being extremely weary about having notetakers on the call, but often wonder either why we use them or how the notes are being used. I think we will see more benefits come from notes as we can better relay candidates background.
Someone Still Has To Own The Workflows
One part of our conversation that I personally don’t hear discussed nearly enough is ownership.
These workflows don’t maintain themselves. Someone has to evaluate integrations, troubleshoot issues, manage APIs, update documentation, improve prompts, and continue refining the process as tools evolve.
Aston owns much of that work for his team today.
It made me wonder whether we’re going to see more recruiting organizations create dedicated AI operations or workflow management roles over the next few years.
Building a workflow is only the beginning. Keeping it valuable is the real job.
My Biggest Takeaway
The engineers I speak with are usually curious about whether recruiters are using AI to source candidates or filter resumes.
Those are certainly use cases.
But when I talk with talent leaders who are actively building, that’s rarely where the conversation stays.
Instead, they’re asking different questions.
How do we walk into kickoff meetings better prepared?
How do we create more consistency across interviews?
How do we reduce repetitive administrative work?
How do we spend more time building relationships with candidates instead of documenting them?
Those are the AI conversations I find the most interesting because they’re focused on improving the recruiting process itself, not just making one step faster.
A big thank you to Aston from PromptMates for taking the time to walk me through how he’s thinking about AI adoption, workflow design, MCPs, and recruiter enablement. Conversations like these continue to shape how I think about where AI is creating the most meaningful impact in talent acquisition.