CodeSignal 发布 Agentic Skills Intelligence，用智能体从真实工作中评估员工技能

The skills intelligence and skills inference market is enormous. Originally pioneered by companies like Eightfold and LinkedIn, early AI companies tried to “infer” skills through your experience, job history, education, and what you did at work. It was an enormous innovation ten years ago and gave birth to companies like Skyhive, Draup, Techwolf, and stalwart SHL.

The original purpose of this technology, all built before GenAI existed, was to help with recruiting. Since recruiting is a $300 billion market (including staffing agencies), this felt like a good opportunity, especially since technology skills were changing so fast. Companies like HackerRank, Codility, and then CodeSignal decided to focus here, building tests and self-diagnostic assessments (highly skilled people train the system on what others need to know), growing quickly.

Then along came Generative AI. The new technology, including ChatGPT, Claude, Gemini, and open source models totally changed the market. We could not only build intelligent assessments but we could also look at actual work artifacts (code, documents, emails, meetings) to deepen this assessment process. Techwolf, which was just sold to SAP, had a number of customers trying to build internal employee assessments this way, often used for redeployment or internal job mobility applications.

CodeSignal, founded Tigran Sloyan, an MIT graduate and former Google engineer, decided to jump into the GenAI opportunity. Unlike many of his competitors, Tigran rapidly built a product called Cosmo, an AI tutor.

I first saw Cosmo several years ago and was somewhat astounded at how it worked. Cosmo could chat with you and based on your needs decide to tive you an interview, test, business simulation, or learning program – all based on your role, job, and experience. In other words, CodeSignal jumped from the single category of “pre-hire assessment” into skills development, personalized learning, and now measuring skills directly from employees’ work.

This has enabled CodeSignal to launch tools for phone screening, individual development planning, corporate training, and even simulation-based skills certification. (We use CodeSignal in our new GHRE Global HR Excellence Certification.) I consider CodeSignal the only platform that uses AI to “certify” skills, not just assess or build them.

This takes CodeSignal into a huge market: the $300 billion recruitment market as well as the $400 billion skills assessment and development market.

This week the company is launching its Agentic Skills Intelligence product, which is a real “development agent” that can watch and analyze your work, assess your skills, and give you coaching and development to improve performance. It relies on a 20+ million data based benchmarking data set and it understands behaviors and real work activities that contribute to your skills, not just the skill itself. Evaluator agents (which are experts) observe your work so CodeSignal gets a full and ongoing profile of your capabilities.

Lots of companies have tried to crack into this market, including the ones above as well as hundreds of smaller vendors with specialized tests, assessments, and development programs. Each tend to succeed by focusing on industries and unique use-cases for their tools. HackerRank focuses on IT assessments. Pluralsight focuses on AI and IT skilling. Eightfold focuses on sourcing and AI interviewing. And SHL focuses on vertical, retail, and sales assessment.

What I find unique and important about Tigran and CodeSignal is the constant focus on the core problem itself: how to identify, categorize, benchmark, and analyze the entire context of a set of skills needed for a particular job. That domain is deep, important, and constantly changing – and CodeSignal’s expansion shows how many applications spawn from the core solution. As we re-engineer almost every job in business, there are a multitude of opportunities to use CodeSignal for real-time skills assessment, categorization, real-time coaching, and learning.

Every major HR software company from Cornerstone to Docebo to Eightfold and most talent acquisition tools have some kind of assessment system, but none are as robust or complete. Major customers include Meta, Anthropic, Instacart, Visa, UnitedHealth Group, Dropbox, Handshake, Mercury Insurance, and The Josh Bersin Company.

We use CodeSignal in our new Global HR Excellence Certification, where the system delivers a real-world simulation and helps HR professionals learn how to solve complex organizational challenges and certify their newly learned skill.

CodeSignal is amazing to see. I recommend you check it out.

Additional Information

What Is Task or Work Intelligence And Can We Decompose Every Job For AI?

The Four “New SoftSkills” We Need To Thrive In The Age of AI (podcast)

Building A Skills-Based Organization: The Exciting But Sober Reality