Job search in the AI era

AI Will Not Find You a Job. It Can Turn Your Search Into a System

Published: 2026-08-13 · Data as of 2026-08-13

Nir Kosover · Founder of JobFox. Building the job-search engine for Israeli tech.

Anyone who has hunted for a job in the past year knows the feeling: dozens of open tabs, applications vanishing into a black hole, and parallel processes blurring into each other. In August 2026, Shahar Polak, Head of Engineering at Imagen and host of the Hebrew engineering podcast “Developers Outside the Box,” published a deck that turns that mess into a method. His starting point is blunt: AI will not find you a job, but it can turn a scattered search into a clear path. We adopted his principles, added context and numbers, and here is the full system in three stages: find the opportunities, get in the door, and prepare for the interview. If you are new to the Israeli market, this is also a map of how hiring actually works here.

Why is searching LinkedIn not enough?

Industry estimates put roughly 70% of open positions off LinkedIn entirely, or on it weeks after they went live on the company’s own careers page. Search only there and you are competing for the roles everyone sees, while missing the ones that have not reached any board yet.

The fix is a wide source map instead of a single source:

1. Company career pages. The official, most current source. A role is almost always live there before it appears anywhere else. As of August 13, 2026, JobFox indexes 15,981 verified open positions scraped directly from the career pages of 5,750 companies in Israel, so you can search all of them in one place.
2. Israeli company directories. Hebrew-language resources like “Mi Megayes Achshav” (“who is hiring now”) list Israeli companies with direct links to career pages, TechMap aggregates tech roles, and Nortech focuses on tech jobs in northern Israel.
3. Focused Google searching. Google Dorking, covered below.
4. Conferences, meetups, and professional communities. The places where opportunities start as a conversation, long before they become a job ad.

How do alternative job titles widen the search?

One role can appear under many names, and searching a single title shrinks the number of opportunities you will ever see. Someone hunting for a Backend Team Lead position should also search:

1. Backend Tech Lead
2. Engineering Manager
3. R&D Team Lead
4. Software Engineering Manager
5. Server-Side Team Lead

Instead of guessing, hand an AI your CV and ask it to derive alternative titles and relevant search terms based only on the experience it contains. Two rules make the output usable: every suggestion must be grounded in what the CV actually says, and the AI may not assume experience, technologies, or domains that are not in it. That builds a wider search map without chasing roles that are not a real fit.

What is Google Dorking and how does it surface hidden jobs?

Google Dorking is precise Google searching with keywords, quotes, and operators, aimed directly inside companies’ recruiting systems. Instead of searching the whole web for “Backend Team Lead,” you search the job pages themselves:

1. site:jobs.lever.co "Backend Team Lead" Israel
2. site:boards.greenhouse.io "Engineering Manager" Tel Aviv
3. site:comeet.com "R&D Team Lead" Israel

That reaches roles published straight on the Greenhouse, Lever, and Comeet boards of Israeli companies. Comeet is worth knowing by name: it is an applicant tracking system especially popular with Israeli startups, so it hosts many local roles you will not see elsewhere. Two habits matter: run the search across all the alternative titles from the previous step, and always confirm the role is still open before investing in an application.

You can also have an AI run these searches for you: ask it to act as a job-search researcher with web access, sweep every title across every tracking system, open and verify each role, and return one table with company, role, location, direct link, and posting date. An honest caveat: manual searching is sometimes faster. AI earns its keep here on scope and consistency, not speed.

Why are conferences and communities a source of unpublished roles?

Conferences and meetups are not just for learning. They are where you meet companies and hear about roles before they hit any board. At most conferences you will find companies with booths, reps, and sometimes members of the actual hiring teams. That is a chance for a real conversation instead of another CV submitted into a system.

A tip most people miss: even if you skip the conference, check who sponsors it. A company paying for sponsorship usually wants to be known, and often it is hiring. The sponsor list of a relevant conference in your field is a ready-made list of companies to research.

Professional communities work on the same principle: roles move between people there before they reach the boards. In Israel much of this happens in Hebrew-language Facebook and WhatsApp groups, such as “Tsarot BeHitech” (“troubles in tech”), “tech jobs before they are published,” and developer WhatsApp groups like DevHelp. For English speakers, olim-oriented groups and local tech meetups fill the same role. Do not join everything. Pick two or three relevant communities and be consistently present.

How do you tailor a CV to every role without rewriting it?

Do not send the same file to every role. The goal is not to invent experience or rewrite who you are. It is to surface, each time, the parts of your real experience most relevant to this specific role.

When working with AI, always attach two things: your current CV and the full job description. Four requests that work, in this order:

1. Fit mapping. Which core requirements of the role already have clear proof in the CV? A table: requirement, proof, strength of fit, worth highlighting.
2. What to highlight, what to trim. Which achievements, projects, or technologies deserve emphasis for this role, and which items can shrink because they are less relevant.
3. Focused changes. Up to five targeted CV edits, each with an explanation of which existing information it is based on. No full rewrites.
4. A recruiter’s read. Ask the AI to act as a recruiter: what gaps, questions, or red flags come up reading this CV? Direct and short, no softening.

One rule repeats through all four: the AI must not invent experience that is not in the CV. If you prefer a structured flow over one-off prompts, dedicated tools do exactly this, including JobFox’s CV studio, which tailors the document to a role and shows you how an applicant tracking system reads it.

How do you drill for interviews with AI?

AI cannot replace your experience, but it is an excellent practice partner. Four uses:

1. Research the company. What it does, who its customers and competitors are, which trends shape its market. Ten minutes of research before an interview changes the quality of the conversation.
2. Build stories from experience. Take a true story about a project you led and ask for help turning it into a short answer in the STAR structure: Situation, Task, Action, Result. That structure turns a generic answer into a clear, specific, credible story.
3. Mock interview with follow-ups. Ask the AI to interview you for the role: one question at a time, wait for your answer, then push with challenging follow-ups. Practicing the follow-ups is where the real value is.
4. Practical round prep. Ask for a realistic case study for the role, propose a solution, then have the AI add a new constraint and probe how you adapt.

A principle worth repeating: build a story bank, do not memorize answers. One good story from your experience serves ten different questions. A memorized answer breaks on the first follow-up.

How do you run several processes in parallel without losing control?

A real job search carries a lot of detail: which roles you applied to, who you spoke with, what was said last, and what happens next. Once several processes run in parallel, organized pipeline management becomes part of the outcome, and memory is not a system.

A simple fix: one tracking board, for example in Notion. Ask an AI to build a workspace with four databases: applications, contacts, tasks and follow-ups, and interviews. For each application track the company, role, job link, status, current stage, contact, last interaction, next action, and follow-up date. Add views for active applications, follow-ups due this week, and upcoming interviews, and close each week with a short review: what moved, what stalled, what needs action.

Who is worth following on LinkedIn?

A large share of what is known about the Israeli hiring market moves through people, not websites. Israeli recruiters and headhunters share insight on hiring, careers, and roles, and some keep job databases that are never published externally. Names worth knowing include Emil Rozenblat, Shahar Shalom, Keren Matudi, Naama Regev, Orin Kaniel, Hagar Israeli, Adar Hagoel, Liat Tal, Batel Rahav, Eyal Siles, Ronit Michlovich, and Mirit Palva Harmash. Beyond following, it is entirely legitimate to reach out and ask whether they have relevant unpublished roles.

Where do you start?

You do not need to adopt the whole system in one day. A sensible first week:

1. Give an AI your CV and build a list of alternative job titles.
2. Run one wide search across all the titles, on the career pages themselves or through an index that already scrapes them.
3. Pick two or three relevant communities and join.
4. Open a tracking board, and from that moment every application gets logged.
5. For every role that interests you, tailor the CV before applying.

Polak’s method comes down to one sentence: AI is not another tool, it is the method that connects the tools. Work this way and you are not searching harder. You are searching organized.