Beyond the CV: Rami Albatal on hiring in the era of tasks

Is the CV still worth reading? Rami Albatal, Beamery's Head of Artificial Intelligence, joined the hosts of the Accelerating AI podcast (Lionel Naidu, Andy Paul and Matthew Rigby) to explain why hiring is moving on from the CV. Along the way: how to keep AI fair, and what the next generation of job seekers needs to know.

Listen to the full episode, "Beyond the CV: AI Recruitment, Fair Hiring and the Future of Education", on Accelerating AI.

From job titles, to skills, to tasks

Hiring has always looked for a shortcut to what someone can do. For decades that shortcut was a job title and a degree. Rami points out that this approach has a built-in problem: it "discriminates against people who didn't manage to go to prominent universities", and "a role title alone doesn't reflect what people know".

Skills-based hiring came next. The idea was to look past the title to what someone knows. But it didn't really take off until recent advances in AI made it practical to pull skills from profiles and match them to job descriptions.

Now the goalposts have moved again. Skills describe what someone can do, but not what they actually do. "I can claim I'm a good software engineer," Rami says. "OK, but what was I doing?" The answer is in the tasks.

"First role titles and educational qualifications, then skills, and now tasks. Now we are in the era of tasks."

When the CV can't be trusted

AI is changing both sides of hiring at once. Candidates use it to write their CVs, and can tailor one for every role in seconds. Recruiters now face thousands of applications per job and use AI to work through them.

One of the hosts put it bluntly: recruiters "can't trust the paper it's written on" any more. Forget what candidates say about themselves. The focus is shifting towards evidence of what they can do.

AI should assist, not decide

With thousands of applicants per role, most organisations will use AI to help sort them. So how do you make sure good candidates aren't unfairly screened out? Rami described three safeguards at Beamery:

• Independent bias audits. An audit partner tests the ranking model against a dataset covering a wide range of industries, ages and ethnic backgrounds, looking for hidden bias.
• Legal involved from the start. "When you design an algorithm or an AI pipeline, go to the legal team as early as you can," Rami says. You don't want to find out at the pre-release stage that a piece of data introduces bias.
• People make the call. The AI shows why candidates appear in the order they do, and recruiters can re-sort and filter in real time.

Rami was clear on the principle behind all three: "In this product, and in any sensitive product, I don't recommend asking the AI to make a decision. The AI should be an assistive tool rather than a decision-making tool."

Transparency is how you comply

On the EU AI Act, most of it comes down to transparency. That means being clear about how you process data and being able to trace every decision. Beamery designs for that. The algorithm can be audited, it explains itself at each step, and decisions are stored so they can be reviewed later.

The same thinking applies to employees' data. People should decide what they share, stop sharing whenever they choose, and ask for their data to be deleted. "Everything is managed by design from the beginning," Rami says.

The hosts noticed. "There aren't a lot of businesses out there taking that stance," one said, calling it "the right way in 2026".

The data is what's hard, not the software

The hosts asked whether AI coding tools could make HR software easy to replace. Rami's answer:

"The thick part is becoming the data: how you curate it, process it and store it in a compliant, scalable way. The SaaS layer on top is becoming thinner and thinner."

Connecting to hundreds of workplace tools, keeping those connections live, and handling storage, deletion and compliance takes real, ongoing effort. That's why large organisations still choose to buy rather than build.

What it means for the next generation

Looking a year or two ahead, Rami sees a clear pattern in software and data roles. The people who openly use AI tools find work fastest. His prediction: "Tech jobs and white-collar jobs will be pretty much about how you manage agents."

But he also argued for keeping the mind working. As AI takes on more of our thinking, he believes education needs to keep "desirable difficulty" until young people are ready to master these tools. He compared it to exercise: "We have technology that means we don't need to move, but we all know we have to exercise to stay healthy. It's the same for challenging our brains."

Watch the full conversation

The episode goes further into world models, robotics and where AI is heading next. It's well worth the 39 minutes.

Watch "Beyond the CV" on Accelerating AI →
Get your free Skills and Task analysis →

Want to see how Beamery helps organisations understand the work their people actually do? Discover Task Intelligence.