Really excited to officially launch AI Sourcer in Elly today!

Last fall, we launched a (very early) beta for the AI Sourcer and over the past few months we rebuilt it from the ground up around one core idea:

Sourcing should feel like continuing the hiring conversation, not rebuilding it from scratch.

Almost every recruiting team still uses LinkedIn, boolean searches, spreadsheets, enrichment tools, and outreach platforms just to run a search. With Elly, you can do it all in one tool:
1. Use Elly as a notetaker on a kickoff call
2. Automatically write a JD, create a hiring plan, and build out scorecards
3. Chat to search across 100M+ US-based candidates
4. Ask Elly ~why~ certain candidates are being surfaced and refine the search as needed
5. Run outbound campaigns on email & LinkedIn
6. Use Elly's ATS to run the rest of your process

Try it out here: https://lnkd.in/eNM3ecMr

Next week, we’re hosting a live “old way vs new way of sourcing” showdown. I’ll be using AI Sourcer while our Head of Search runs a traditional LinkedIn + boolean workflow side-by-side on the same role.

Register at the link in the comments!

The amount I use AI for over the last few months has increased exponentially, yet I still class myself as user rather than truly adopting AI.


Today i've set up a fitness and protein tracker for the week.
I've used it to help with some garden design.
It's helped me write a basic advert for an AI Software role.
It's helped me build a 6 week football coaching plan.
It's been my sound board for a new management role i'm expecting in.
It's even helped me build the layers to make canva look like an old amber CRT.


The one thing it doesnt help with is my content - not because it wouldnt be good at ti - it would definitely get better engagement than the stuff I create... But the people piece is mine - i'm not handing over my writing, my resourcing, my profiling to a machine - when I do that then wheres the achievement?

When a post does well and generates conversation - thats on me!
When I find someone their perfect career - thats my contribution!
When I make a mistake - its mine to learn from!


AI is now my administrator andits better and faster at lots of the basics than I can ever claim to be - a lot of the stuff I just wouldnt have bothered doing before.


The people i'm hunting for though - are the AI Natives - the engineers who have agents writing code and building products, all orchestrated by their skills and experience - the managers who have teams building with AI and are managing both the humans and the controls and security around AI generation - these people still blow my mind!

AI tools are not replacing recruiters. They’re replacing bad recruiters.

I use AI every day in my process.

For:

✔ Initial CV screening patterns
✔ Job description optimisation
✔ Interview question generation
✔ Market salary data synthesis

Not for:

✘ Deciding if a person is right for a team
✘ Reading between the lines of a candidate’s career story
✘ Managing a hiring manager’s expectations
✘ Negotiating an offer that both sides feel good about

The recruiters panicking about AI are the ones whose entire value was in admin.

The recruiters thriving right now are the ones whose value was always in judgment, relationships, and insight.

And if you’re a hiring manager — are you working with a recruiter or a CV forwarding service?

Most recruiters are using AI wrong. Heck, I was using AI wrong in the beginning. They treat it like Google.

“Write me a job description.”
“Summarize this resume.”
“Draft an email.”

That’s not where the value is.
The real shift in thinking is this...
AI is not a tool but it’s a workflow layer.

Here’s what that looks like in practice...

Before AI:
1) Open req
2) Write JD from scratch
3) Manually screen 50 resumes
4) Start sourcing

After AI:
1) AI drafts the JD from intake notes
2) AI summarizes and ranks applicants against non-negotiables on the req
3) AI generates search strings
4) Recruiter focuses on decisions, not typing

Same job. Completely different leverage.

The teams who will be the most efficient won’t be the ones “using AI.”

They’ll be the ones redesigning their workflow around it. This doesn't need to be scary. It's actually super cool.

I’ve been heads-down building my AI workflow for the past few months.

As a Talent Partner at a VC fund, I’m constantly context switching: sourcing candidates, prepping for calls, writing content, tracking market signals, supporting founders.

The cognitive load is real.

So I started systematically automating the parts that don’t need me to be me so I can spend more time on the parts that do. 🛠️

Here’s what I’ve built so far.

🔍 Talent workflows
• Auto generated weekly prep sheets for all candidate 1:1s pulled from calendar and intake notes
• “Jobs of the Week” LinkedIn posts drafted automatically from role inputs
• Candidate matcher that scores my entire talent pool against any role in seconds
• LinkedIn network matcher using the same scoring across my 20,000+ connections
• Student finder that analyzes project depth not just keywords to surface top interns and new grads 🎯

📊 Market intelligence
• Daily briefing with layoffs, acquisitions, funding rounds and ready to use Boolean strings
• Weekly digest that turns signals into a portfolio ready email
• Monthly report covering news and press for every portfolio company 📰

🧠 Founder support
• Values based interview guide generator customized in minutes
• Talent playbook builder with Notion ready guides for hiring and people topics from pre seed to Series A
• Daily briefing also includes AI news and top stories

💡 Bonus
On workout days it tells me what to do at the gym. Yes, really. 💪

I’m not a developer.

I built all of this in Claude.

Took a few weeks of iteration. Saving hours per week already.

If you’re in recruiting, VC, or any role heavy on relationships and content, building your own system is very worth it.

The ROI is real.

Here's 5 recruiting agents I use daily at Abnormal AI.

🔍 Outbound Sourcer: Thinks like a top recruiter and a top candidate at the same time. Reads the profile, infers what motivates them, writes outreach that actually lands. Not templates. Actual personalization at scale.
📝 Intake Agent: Stops bad reqs before they open. Pushes back on vague job descriptions and forces clarity on what "good" actually looks like. Less interview noise. Fewer late-stage surprises.
🎯 Debrief Synthesizer: Turns messy panel feedback into a clear hiring decision. Flags conflicting signals, catches bias patterns, and makes a recommendation. Hire, no hire, and why.
🦄 Recruiting Ops Automator: Finds where your process is quietly breaking. Time-in-stage, funnel drop-off, and offer slowdowns — surfaced automatically, no manual report needed.
🏆 Talent Brand Agent: Keeps your voice consistent across every candidate touchpoint. Same positioning, same tone, no drift. It also includes employee stories for the different job families.

The numbers across two quarters: 88% offer acceptance with a 46.6% InMail response rate — nearly 2x benchmark, up 22.6% since October.

That's what recruiting on systems looks like. Most teams have the same tools.

The gap is whether you're building or waiting.

I keep building things and then not knowing how to talk about them.

Not because they're complicated. Because every time I sit down to write a LinkedIn post, I realize the thing I want to say is actually 1,200 words long :)

So I'm putting it all in a new blog. I want us to share what we're building and to show that getting there isn't nearly as intimidating as it sounds. I started where most people start. I made a lot of mistakes. I lost track of time in my home office more than once. And I got somewhere I couldn't have imagined a year ago.

The first post is about the last year: starting with ChatGPT and job descriptions, learning to connect to data, joining Formation Bio, and eventually building tools with Claude Code that I use every single day that have fundamentally changed how I work.

Special shout out to those I've met along the way. Jason Miller , Joe Atkinson and the entire PromptMates community - it's been so fun to learn how to do this in talent together! If you're curious what that actually looks like inside a company that takes AI seriously, I hope you'll read it. https://lnkd.in/eiGPccRK

I've been using Claude for a while now, and one question keeps coming up - will AI take over recruitment? 🤔

Honestly, it's a fair question. AI can parse CVs, write job ads, summarises interviews, and research companies faster than any of us.

So... are we done?

Not even close. And here's why.

Recruitment is one of the most human industries there is.

You're not just matching skills to a job spec. You're figuring out why someone really wants to leave their job. Picking up on the nerves in a candidate's voice. Knowing when a client says "culture fit" they actually mean something else entirely. Reading the room on a difficult conversation.

No algorithm does that.
That's not a skill AI has. That's a people skill.

The recruiters who'll get left behind aren't being replaced by AI, they just stopped picking up the phone.

Skynet can have the admin. The relationships are ours. 💪 🤖

#AIinRecruitment #FutureOfWork #AI #ArtificialIntelligence #Recruitment #TalentAcquisition

been in HR for 6 years and just got access to Claude for our team, trying to figure out what's actually useful vs. what's just hype, so far I've used it to: draft job descriptions (saves me like 30 min each), summarize exit interview notes into themes, help managers phrase tricky feedback emails

that said feel like im barely scratching the surface, what are you all actually automating or streamlining with AI tools? specifically curious about:
- anything that's saved you legit hours per week
- tasks you WISH you could automate but haven't figured out how
- things that flopped/weren't worth it

not looking to replace the human part of my job, just trying to spend less time on the tedious stuff so I can actually talk to people ... tips appreciated

When I first used AI to draft job descriptions and posting ads with conducting search and replying email its instant output left me thrilled yet anxious—amazed by its efficiency, but fearing #TalentAcquisition or HR generalist might be replaced by this invisible tool. As I work with AI daily, I’ve gained a clear view of its pros and cons, and learned how to strengthen myself amid this shift. Such as AI-enabled Moka HR SaaS or LinkedIn redefinment of quality and quantity of output; embrace the power of AI adoption.

AI brings tangible benefits, especially for operational and administrative level. It quickly supplies materials and coherent drafts, acting as my inspiration assistant to free me from repetitive work like researching and outlining. This lets me focus on deep and boosting both quality and productivity. For data analysis, AI sorts messy data, spots trends and anomalies fast, cutting hours of late-night work and speeding up decisions. It handles tedious, mechanical tasks, lifting my efficiency greatly and leaving time for high-value thinking.

Yet AI’s flaws weigh on me and #Everyone. Over-reliance has dulled colleagues’ writing skills, stripping their work of soul until they were replaced. AI also spreads false information confidently, so careless use leads to mistakes. More worrying, it’s replacing routine jobs like simple data analysis, leaving me anxious about my own security.

To stay #irreplaceable, I now craft emotional, user-centric content that AI cannot replicate, and dig into the business logic behind data instead of only trusting AI results. I’m learning video editing and digital operations to diversify my skills, and mastering advanced AI prompts to turn it from an assistant into a partner.

In this fast-evolving AI era, I keep learning like a sponge, follow industry updates, and build cross-domain integration skills to create unique value. AI is opportunity and challenge alike—I will not be over-optimistic or fearful, but keep growing and forge my own way in this changing world.

“The AI Tsunami Is Unstoppable — Here’s How to Ride the Wave. Cultivate Relentless Curiosity. The wave is here. Don’t just survive it—master it”