Child & Youth Services

Child & Youth Services covers nonresidential individual and family social services for children and youth — adoption services, child welfare, family preservation, foster-family training, juvenile-delinquent prevention, and youth crisis intervention. Services are delivered by a mix of public-sector child-welfare agencies (state DCFS departments), large nonprofits (Annie E. Casey Foundation, Casey Family Programs, KVC Health Systems), and a long tail of community-based providers. Total US spend on child-welfare and youth services exceeds $30 billion annually. The workforce is small but high-stakes — roughly 35,000 child-welfare caseworkers nationally with annual turnover of 15-25 percent driven by caseload weight, vicarious trauma, and pay levels well below private-sector social-work rates. Caseload sizes typically run 15-30 cases per worker, often two to three times recommended ratios. Information systems vary wildly by state — some still run ancient mainframe SACWIS systems while others have modernized in the past decade.

Casebook is a leading case-management platform purpose-built for child-welfare and human-services agencies — supports caseworkers across child welfare, family preservation, and foster-care work with AI-assisted documentation, narrative summarization, and outcomes tracking. Bonterra (formerly Apricot and Social Solutions) is the dominant case-management and impact platform across nonprofits and government child-and-youth agencies, with growing AI features for program reporting and donor engagement. Predictive risk modeling has been deployed by some jurisdictions — Allegheny County's child-screening tool is the most-studied example — with sharp policy debate over fairness, transparency, and racial bias. Population-health platforms (Innovaccer, Palantir AIP) extend into the public-sector social-services space for some state agencies. The horizontal foundation models (ChatGPT, Claude, Microsoft Copilot, Perplexity) have rapidly become daily aids for caseworkers — drafting court reports, summarizing parent narratives, translating between languages in family meetings, and accelerating the documentation work that dominates caseworker time.

The biggest opportunity is documentation relief. Caseworkers routinely spend 50-70 percent of their time on case notes, court reports, and home-visit narratives — work where AI assistance could meaningfully reduce burnout and let workers focus on client-facing time. The biggest risk is predictive-risk modeling. Tools that score families for child-removal risk are among the most ethically fraught AI deployments anywhere — false-positive rates have direct, immediate harm (family separation), and historic data carries deep racial and class bias. Major academic work has put pressure on jurisdictions to require transparency, audits, and human oversight before deploying these models. The platforms that will win in this space are the ones that lead with documentation assistance and family-engagement tools (low risk, high benefit) rather than predictive-removal models (high risk, contested benefit). Federal funding is flowing toward modernization of state SACWIS systems, with significant opportunity for AI-augmented case management to reach much of the country.