CodeSignal Launches Agentic Skills Intelligence to Evaluate Skills From Everyday Work

A company can know who has completed a course, adopted an AI assistant or held a particular job title without knowing how well that person can perform the work ahead. CodeSignal is targeting that gap with Agentic Skills Intelligence, a new offering that evaluates employees’ work and connects the resulting skills profiles to personalized practice and coaching.

Announced as generally available on October 28, the platform uses specialized agents to evaluate work from systems employees already use, rather than relying on self-ratings or keywords in a résumé. CodeSignal says the goal is to give leaders a continuously updated picture of workforce proficiency and a way to address the gaps that picture reveals.

The launch extends the company’s assessment and learning business into the flow of everyday work. Its potential value lies in connecting three activities that often remain separate: collecting evidence of capability, deciding where development is needed and helping employees improve.

Evaluating Work Against a Structured Skills Rubric

CodeSignal’s announcement names systems including Gong, Salesforce, GitHub, Cursor, Claude, ChatGPT, Workday, Greenhouse and Google Workspace as sources of work its agents can evaluate. The work is scored against structured rubrics grounded in industrial-organizational psychology, the discipline concerned with behavior and performance in workplaces.

The company positions that process as a more direct measure of proficiency than inferring skills from a title, performance review or completed course. An employee’s output becomes evidence for a skills profile that can be updated as new work becomes available.

CodeSignal cites more than 21 million skill evaluations completed across its platform as background for the assessment approach. That is a cumulative platform figure, rather than the number of employees already evaluated by this newly announced product.

The distinction between evidence and inference is useful, but it does not remove the need for judgment. A work sample must actually reveal the skill being scored, and the rubric must suit the task. A strong result produced with AI assistance, for example, raises questions about which capabilities were demonstrated: writing an initial solution, directing an assistant, checking its output or combining those activities effectively.

That makes the underlying assessment criteria as important as the number of connected systems. More workplace data can expand coverage, but it does not automatically make every skill visible or establish how much an individual contributed to a collaborative result.

Connecting Skills Gaps to Personalized Practice

The development side of Agentic Skills Intelligence uses Cosmo, CodeSignal’s AI tutor, to provide contextualized practice. The proposed loop is to identify a proficiency gap, give the employee relevant opportunities to practice and then track whether their skills improve.

CodeSignal’s existing Learn Academies provide context for that approach. They combine courses, hands-on practice and embedded assessments into guided programs, with Cosmo offering feedback during learning. The emphasis is on applying a skill and checking progress along the way.

The company’s business courses extend practice beyond programming to activities such as customer conversations, writing tasks and financial analysis. That broader scope matters for a workforce product: skills development may involve practicing a sales pitch or exercising managerial judgment as well as solving a technical problem.

A sales leader, for instance, might use the new system to identify where a team needs help presenting a product, then direct tailored practice to those employees. CodeSignal offers this as a use case for the platform, not as a disclosed customer outcome from the launch.

For employees, the useful result would be a specific development path. A score is more actionable when it explains what needs improvement and leads to an exercise that addresses it. The product’s effectiveness will depend on how well that connection holds across different roles and types of work.

Measuring AI Capability Beyond Adoption Counts

CodeSignal is also framing the product as a way to examine whether AI investment translates into better work. Its announcement cites company research in which 84% of organizations expanded AI software investment and 79% increased AI training over the past year, while 79% could not measure the return on that spending. It also reports that six in 10 managers make critical talent decisions using guesswork and 66% of employees say their skills have never been formally assessed.

Those are findings reported by CodeSignal. The supplied release does not provide the survey’s sample size or methodology, so the figures should be read as context for the company’s product thesis rather than a comprehensive measure of all employers.

The practical question is nevertheless clear. Tracking assistant usage can show that employees have adopted a tool; it says less about whether they can use it effectively. Evaluating work against relevant criteria could help employers distinguish frequent usage from stronger proficiency.

That would also create a more meaningful basis for development spending than counting course completions alone. A measured improvement in a relevant skill is a useful signal, although translating that improvement into business return still requires attention to the role, project and other factors affecting performance.

Bringing Skills Insights Into Existing Workflows

CodeSignal says its MCP server lets teams ask questions about skills and jobs data from an AI assistant of their choice. Model Context Protocol provides a common interface for connecting an assistant to external tools and information; here, the intention is to make workforce insights accessible without requiring users to leave their usual assistant.

The announcement also includes organizational skills-gap analysis, workforce planning and market intelligence on hiring demand, roles and pay bands. Those are related but distinct information sources: external labor-market trends describe demand beyond the company, while employee profiles are intended to describe demonstrated capability within it.

Existing CodeSignal skills intelligence materials describe benchmarking, multi-level analytics and skills acquisition tracking. The new launch adds the emphasis on continuously evaluating work and linking the resulting evidence to individual development.

The Importance of Context in Workforce Decisions

More current skills information could help a leader find internal expertise for an urgent project, identify a development need or plan where recruiting is necessary. It also changes the significance of everyday work artifacts: material produced to complete a task may become evidence used to assess an employee’s proficiency.

That makes clarity about the evaluation process important. Employees and managers need to understand which work informs a profile, what the rubric measures and where the evidence is incomplete. Limited exposure to a task should not be confused with demonstrated inability, and an automated score should be considered alongside the context in which the work was performed.

CodeSignal’s launch brings assessment, workplace evidence and AI tutoring into a closer feedback loop. The opportunity is to make workforce development more specific and responsive. The test is whether that loop produces useful, explainable evidence of skill and helps employees build the capabilities their work actually requires.