Previously you were a serial entrepreneur, acting as founder & CEO of several startups. What were some of the biggest hiring challenges that you encountered?
Hiring has been one of the most challenging aspects of my entrepreneurship journey. As entrepreneurs, we know people matter more than anything else and building the right team is the single most important job of any business leader. However, it is really tough to allocate the sufficient amount of time needed to find the right people when you’re maintaining so many other business activities involved in starting and scaling a company. Without objective data on who is available out there, it is hard to find the right set of people, and even harder to know if they will do well in your organization.
Could you share the vision for how Findem is building an autonomous talent platform for the HR team of the future?
Talent acquisition is a complex job with hundreds of tasks, done by tens of personas, across tens of point tools that do not talk to each other most of the time. Our vision is to remove this complexity through a combination of AI and workflow automation.
Our first and foremost goal is to support the talent teams by automating away mundane, repeatable and error-prone tasks from their day-to-day and assist people in making faster, better and more fair decisions with data. We’re already seeing use cases, such as a large tech company where they were using eight to 10 systems just to build a talent pipeline, and each was used in a siloed manner. It was taking them 80-100 clicks to accomplish a single task and now, with autonomous applications, they can perform the same task with one click.
Like nearly all business functions, talent organizations will undergo an AI-first transformation and our plan is to automate everything that can be automated, enabling recruiters and other talent professionals to reach their fullest potential. Autonomous applications will initially play a pivotal role in planning, pipeline and analytics, and then extend across the entire talent lifecycle, encompassing everything from workforce planning to talent pools to career development and succession planning.
Findem analyzes trillion of data points and takes advantage of what is called 3D data, could you clarify what 3D data is?
Findem ingests 1.6 trillion data points from hundreds of thousands of sources to generate entirely new talent data that doesn’t exist anywhere else and provides an understanding of an individual and the companies they’re associated with, over time. Findem uses these three dimensions of data – people and company data over time – to connect individual and company journeys and create enriched talent profiles.
Think of it this way: every person who’s worked in the modern job market has a journey and they leave behind a digital footprint. There are titles, job promotions, certificates, code contributions, publications, social posts and so forth. Similarly, companies have a journey. They have activities such as rounds of funding, IPOs and financial filings, as well as job descriptions, org charts, company reviews and leadership profiles – all of this data can chart an organization’s development and progress.
Traditionally, talent decisions have relied on a resume, job application and/or LinkedIn profile that only offer a one-dimensional slice of a person and company data. However, we’ve built a platform that’s capable of capturing thousands of data-points on people and company journeys and converting them into a massively enriched profile. The result is a more detailed and granular understanding of a person’s experience, skillset and impact than what was previously possible with manual research or from a user-generated LinkedIn profile.
With our Talent Data Cloud, entire careers are searchable on command through a GenAI interface. For example, you can ask the platform to show you CFOs at U.S. companies owned by PE firms who took a company from a negative to a positive operating margin or to give you a list of loyal product managers who worked for a B2B startup and saw it through a large Series C.
What are the different types of data points that are analyzed?
Our Talent Data Cloud dynamically and continuously leverages a language model to generate 3D data from hundreds of thousands of data sources.
It analyzes profile and contact data from the likes of LinkedIn, GitHub, StackOverflow, Kaggle, Dribble, Doximity, ResearchGate, WordPress and personal websites. Census data comes from the U.S. Census Bureau, of course. Additionally, we look at company data from funding announcements, IPO details, business models of over 8 million companies, and over 100,000 aggregated company and product categories. For verified skills, the platform analyzes over 300 million patents and publications, over 5 million open dataset and ML projects, and over 200 million open-source code repositories and other public contributions. And we importantly include ATS data that includes applicant profile information from the user’s ATS, which could be Greenhouse, Workday, SmartRecruiters, BambooHR, Lever and so on.
What is machine learning looking for when analyzing this data?
Findem is BI first, then uses AI to learn and make predictions based on factual data. We call this a deterministic model vs. a probabilistic model. For instance, we do not probabilistically infer that you have startup experience, we instead look at your employment history and see if any companies you work at have been classified as startups and then add a ‘startup experience’ attribute against your profile.
How is this data then transformed into attributes, and what are attributes?
Once data collection happens, we have an intelligence engine (think of it as a sophisticated SQL middleware) that can map data to any attribute we would like to create.
Attributes are the skills, experiences and characteristics of individuals and companies – and they’re both tangible and intangible. Tangible attributes include roles (current, past and role experiences), work experience, education, qualifications and other technical information. Intangible attributes can be far reaching, such as whether someone inspires loyalty, builds diverse teams or is mission driven.
Our attribute-based search enables HR teams to search for candidates across all channels in their talent ecosystem using practically any criteria you can think of.
How does the platform prevent gender or racial AI bias from creeping into hiring decisions?
Our platform was intentionally designed to not make decisions on behalf of any user, but rather for AI to assist the people in their decision-making. Using a BI-first strategy, the platform prioritizes the collection, analysis and presentation of data to provide insight and support for decision-making, then uses AI to learn, reason and make predictions or recommendations with trusted outcomes.
We’re a searching and matching platform, not a candidate evaluation platform, and AI is never used to make a subjective evaluation of a person. It never automatically advances or rejects applicants. Also, since Findem doesn’t use AI for searching and matching (these capabilities are BI based), it mitigates the risk of bias or discrimination creeping into the process.
How does Findem simplify the process of promoting internal staff?
At the core of it, we do not have to differentiate between ‘internal’ and ‘external’ talent. For any person in our database, our algorithm can find top-matching candidates whether they are outside or inside the organization.
What are all of the talent management tools that are offered?
We are consolidating top-of-funnel activities, so everything from talent sourcing to CRM to analytics. We also have a solution for internal mobility and we’re rolling out offerings for referral management and succession planning.
At what stage of the entrepreneurial journey should a startup be at before they reach out to Findem?
We service customers of all sizes, but our sweet spot tends to be companies that are in scaling mode with a few hundred employees.
Thank you for the great interview, readers who wish to learn more should visit Findem.
Credit: Source link
