Hiring Assistant 2 starts reaching recruiters this month. LinkedIn says that for global customers who already use Hiring Assistant in English, it will "begin to automatically update with new capabilities for free beginning in November" (LinkedIn News).
We covered the trust side of this update back in October. This post is about something more practical. The biggest change in version 2 is memory. The assistant now learns how you and your company hire, and it uses that to fill in the gaps when your search is thin. That can save a lot of time. It can also lock in habits you never meant to keep.
So before it starts learning from you, decide what you want it to learn.
What the memory actually does
ERE's write-up of the Talent Connect keynotes explains it well. Hiring Assistant 2 has "a persistent memory tied to each recruiter" and fills in a sparse search from the recruiter's and company's hiring history (ERE).
LinkedIn's own example: you ask for a product manager, and it returns people near the office, in retail, at mid-level, because that is who you have hired before.
Read that example twice. If your last five product managers came from retail, the assistant assumes the sixth should too. That is fine if retail experience is what the role needs. It is a problem if the role has changed and nobody told the tool.
ERE also notes that recruiters can review and confirm the criteria the assistant infers. That review step is where you get control back. Use it every time.
Three things to settle before the update lands
1. A written scorecard for each open role
The assistant can read loose phrases. ERE says you can type something like "fresh perspective" and it will try to interpret it. But "fresh perspective" means different things to different hiring managers. If you don't pin it down, the tool will fill in the meaning from your past hires.
Before you open a new search in version 2, write down:
- The three to five things this person must be able to do in the first six months
- What would rule someone out, in plain words
- What is nice to have but should never decide a shortlist
Keep it short. You want something a hiring manager can agree to in one meeting. Then, when the assistant shows you its inferred criteria, compare them against this list. Anything on its list that isn't on yours should be a question, not a default.
2. Your hard no-go rules
ERE reports that Hiring Assistant 2 "can also apply company rules, such as never recruiting from the company's own subsidiaries."
That is useful, but only if the rules exist somewhere clear. Most teams carry these in people's heads. Write them down before the rollout:
- Companies you won't recruit from (subsidiaries, close partners, clients with agreements)
- Locations you can't hire in for legal or payroll reasons
- Any role-specific limits, such as needing a certain work permit
If a rule is fuzzy for your own team, it will be fuzzy for the assistant too.
3. What "fit" means at your company
Version 2 adds two signals to each recommended candidate. Fit signals give a personalized read on how well someone matches the role. Candidate interest estimates how open someone is likely to be, using platform engagement, Open to Work details and how often they reply to InMail (per ERE).
LinkedIn says fit signals help hirers understand who matches the role and also who may be a strong fit for their company, and why (LinkedIn News).
"Fit for the company" is the phrase to be careful with. It can mean shared values and a working style that suits the team. It can also slide into "looks like the people we already have." Agree internally on what fit means for you, and make sure it is about the work, not about background.
One good sign from ERE's example: fit signals can flag a mismatch, like a candidate who grows existing accounts when the role needs someone to find new ones. That is the kind of fit read you want. It is tied to the job.
How to tell when it is copying your past instead of filling the role
Memory works on history. History is useful until the role changes. Here are a few signs the assistant is matching old hires rather than the job you have today.
Every shortlist looks the same. Same industries, same company sizes, same career paths. If the role is new or has changed, a shortlist that looks exactly like your last three hires deserves a second look.
Criteria you never typed keep showing up. Location, seniority or industry filters appearing on their own are a clue the tool is filling gaps from history. Check them against your scorecard.
Strong candidates from other backgrounds rarely appear. If you know good people exist in a different industry and they never surface, try running the search with clearer, written criteria and see what changes.
The hiring manager keeps saying "not quite." When the people you send match the old version of the role, the manager will feel it before you can explain it. That is often the moment to reset the brief.
A simple habit helps here. For any role that has changed since you last filled it, start the search by telling the assistant what is different this time. Don't assume it knows.
Use the speed, but keep the judgment
LinkedIn shares some strong numbers for the first version of Hiring Assistant. According to its announcement, recruiters are 4x more likely to contact candidates the assistant sources, they review 83% fewer profiles to find qualified matches, they find interview-quality applicants 33% faster, and prescreening is completed in a median of 6 minutes (LinkedIn News). Those are LinkedIn's figures about its own product, so read them that way.
Version 2 also reaches further into the process. ERE says it moves into screening and scheduling, and for teams that connect an ATS (it lists Greenhouse, Lever, SmartRecruiters, Zoho Recruit and Tracker), it can evaluate the whole talent pool, including people who applied outside LinkedIn. When you move a candidate forward in LinkedIn, the change syncs back to the ATS.
That reach is the reason to set rules first. A tool that touches screening, scheduling and your ATS is making more small decisions on your behalf. Small decisions based on old habits add up quickly.
A short checklist for this month
- Write a one-page scorecard for each open role before you search
- List your no-go companies, locations and permit rules in one place
- Agree with hiring managers on what "fit" means, in terms of the work
- Review the assistant's inferred criteria on every search, including the tenth one
- For changed roles, tell the assistant what is different before it fills the gaps
- Look at your shortlists every couple of weeks and ask if they all look the same
The update is free and automatic. The thinking behind it is still your job. Spend an hour on these rules now, and the memory will learn the version of your hiring you actually want.
