Introduction

Hello everyone, I'm Luca Vecchio, founder and CEO of Machiguru, and Omar has spoiled a little bit our story, which is, now I don't have the surprise effect, but don't worry. So, it's a simple story, okay? We are startuppers, we are innovators, so we think that with technology we can solve all the problems, no?

The Talent Hunting Problem

The problem that we want to solve is the talent hunting problem. It's a really complex problem because the process is really long. It's a really long process and it involves many actors, companies, candidates, recruiters, different interviewers, different tests.

So it's a really unstructured process process with a really big amount of unstructured data. What we try to do?

Attempt 1: Data-Driven Matching

We try to gather all the data together, so all the data from candidates, so salary expectation, ideal environment and cultural assessment, technical skills and location, so we gather all the data from the candidates.

We gather all the data from the company and we try it with ex -Amazon data other scientists, that kind of level of force to build a model that can help to match candidates and companies with the right results.

When the Model Failed

The results were not good because, as Omer told you, we made zero placement in one year with this test, but we discovered our strength.

1Our strength was that all the clients that we have were still asking us to help them them to discover better their needs because the problem is that they don't have the problem to find the right candidate or to find the right expert but to define well the needs that the company has.

The Real Issue: Misaligned (and Inaccurate) Inputs

So, in the beginning, from the beginning to the end, the problem is one that, sorry, I went a little bit over, you messed up my presentation, the problem is one, the model is not working because the people are lying. because everybody lies. What it means?

The candidates are lying in the questionnaires and are lying during the interviews, so the model is not working because it's proposing them different opportunities compared to to the opportunities that they really want.

On the company side, the companies are lying, as you can see on all the job postings, are lying on the budget, are lying on the clear idea of the role and the activities, and they're lying also on cultural expectations.

I'm talking about things, yes, we have a flexible environment, but don't mean many times that it's a real real flexible environment and things like this.

So, in the end, we were quite messed

up with this problem, but we were still working as a classical recruitment agency, bringing the business moving on and trying to understand how to scale the business and how to solve

Pivot: Building an Independent Talent Network

this problem at the beginning at the beginning of the last year one investor told us what we will what you will suggest to a friend that has to find someone for his company with the best possible rate

between time quality and cost and at this point we i will be a little bit uh proud of myself but my and our thinking was us not as much guru but as an independent and enter which lower cost between a great firm and higher quality the higher quality work

Why Off-Platform Candidates Matter

because the main problem is that the real competitive advantage in recruiting market is not having the candidate on LinkedIn, having the candidate on Indeed, but it's having the candidate which is not on the platforms, because the

company are really trying to find those candidates, because our candidate with less competition between the candidates that are already on platform, that they have already access, because all the data from LinkedIn, Indeed, blah, blah, blah, it's already public, so the companies are struggling to find candidates outside

the common pool which probably are also the best candidates because are not looking for a job so we think that is a strong competitive advantage and we put all these candidates together with an independent network of it and that's at this moment we are the quite first and bigger network of it hunters in Italy

and we are growing up this model so if I'm we are not an AI company we have not an AI working model, we are not cutting -edge technologies.

How Machiguru Uses AI in Practice

Why am I here talking about this story?

Automating Admin and Repetitive Work

Because we are still using AI to solve this incredible complex process. Because there are still a lot of full, it's a long process made of full admin and repetitive activities that are making our recruiter or recruiters lost quite important amount of time during the months.

So what we are doing? We are

executing and eliminating simple admin tasks with CSV optimizations

that we are suggesting to the candidates at our platforms, drafting JDs with with GNI, with a fine -tuned model which helps us to do it better.

And reading and storing tons of CVs in an immediate manner.

Search and Matching to Speed Up Shortlists

Then another level, we are searching for new candidates. So matching candidates with position with an that helps.

Our is opening the platform, putting a candidate, and that's already four to five positions that are already fitting for that candidate and then can call the candidate interview the candidate and discover if it's a real fit so not presenting the candidate if

we are not true of the real fit then we suggest our recruiter possible candidates for each new positions what means we put a new position on the platform perfect on based on the candidates that we already have on the database and on the network or the recruiters we suggest the best 5 fitting positions for that, sorry, the best 5 fitting candidates for that position.

That means in this way we can cut off two weeks of work because probably if we already intercepted that candidate we already know that it's a fit so we can interview it and present it to the company.

In the end

Candidate Communication and Interview Data

And we have to meet the communications with candidates, which means, I don't know if you were already involved in a recruiting process, but many times there is a good amount of admin

and caring activities, like a recruiter which is reading to you in the morning, hello, remember that you have an interview today, or a feedback after the interview, or I will help you to getting you the offer in the end there are a good amount of activities that we

automate with our platform so in the end all these activities are bringing to create a huge amount of data that we need to manage and we track all the information from the interviews all the questions from the interviews and all the incoherences between what candidate wants and the JDs that you gave and we

We are trying, in this moment, to storage all this data in a compliant way with GDPR and other European normations. And we are trying to automate more these things.

Toward AI Agents in Decision Support

What means that probably we are still trying to build many AI agents that are automating small activities also on the decisional part.

Like, you are in an interview, and the agent is helping you, like, suggesting new questions that you don't already made to the candidate and that's all so with with

Results: More Clients per Account Manager

these activities at this moment on this moment there are only six months that we are in this in this pure we are already helping four times the clients we are

helping before so this moment if you ask to a classic consultant in a traditional film you will tell you I manage ten clients we our account manager is is managing at this moment.

I wrote Forex.

In reality, we are managing 15 clients per account manager.

And we are helping our red hunters to not have any admin activities, and they can still focus only on discovery and selecting the right candidates and nurturing relationships

with their networks. That's all.

Conclusion

So in our contest, what happened? 1And AI simply makes the same problem, but faster. So we discovered faster that it was not the right choice. OK?

Thank you. Thank you, Luca.