NOR-001

NAV "Frida" AI Chatbot for Social Benefits

Download PDF
Norway Europe & Central Asia High income Full Production Deployment Confirmed

Norwegian Labour and Welfare Administration (NAV / Arbeids- og velferdsetaten), operating under the Ministry of Labour and Social Inclusion. Technology provided by Boost.ai (Norwegian conversational AI company). (Boost.ai; University of Agder research)

At a Glance

What it does Perception and extraction from unstructured inputs — User communication and interaction
Who runs it Norwegian Labour and Welfare Administration (NAV / Arbeids- og velferdsetaten), operating under the Ministry of Labour and Social Inclusion. Technology provided by Boost.ai (Norwegian conversational AI company). (Boost.ai; University of Agder research)
Programme NAV Multi-Channel Service Delivery — Frida Virtual Agent
Confidence Confirmed
Deployment Status Full Production Deployment
Key Risks Operational and system integration risks
Key Outcomes During COVID-19, Frida answered over 270,000 citizen inquiries within weeks, with approximately 80% of interactions resolved without escalation to human advisors.
Source Quality 3 sources — Academic journal article, Other

"Frida" is a conversational AI chatbot deployed by the Norwegian Labour and Welfare Administration (NAV — Arbeids- og velferdsetaten) to provide 24/7 automated assistance to citizens accessing information about Norway's social benefit programmes. NAV is Norway's principal public welfare agency, responsible for administering approximately one third of the national budget through programmes including unemployment benefits, sickness benefits, pensions, child support (including the cash-for-care benefit), disability benefits, work assessment allowances, and other social insurance and social assistance schemes. Frida was initially launched in 2018 and was already operational when the COVID-19 pandemic struck Norway in March 2020, at which point it proved critical to managing a massive surge in citizen inquiries.

Frida is built on the Boost.ai conversational AI platform, a Norwegian-developed enterprise virtual agent solution. The system uses natural language processing (NLP) to understand citizen inquiries submitted via text chat on nav.no and provides automated responses drawing on NAV's knowledge base of policies, procedures, eligibility criteria, and service information. The chatbot is designed to handle a wide range of topics across NAV's service portfolio, providing general information about benefits and services, guiding citizens to relevant application forms and digital self-service tools, and answering frequently asked questions about eligibility, processing times, and documentation requirements.

The system operates with a structured escalation pathway. Frida handles initial citizen interactions autonomously, attempting to resolve inquiries through its automated knowledge base. When the chatbot cannot resolve an inquiry — either because the question falls outside its trained topics, requires access to personal case information, or involves complex circumstances — it can transfer the conversation to a human advisor at NAV's contact centre. During weekday business hours (9:00 to 15:00), citizens can request to be transferred to a human advisor through the chatbot interface. Outside business hours, Frida operates as a standalone 24/7 information service.

Frida's significance was dramatically demonstrated during the COVID-19 pandemic. When Norway entered lockdown on 12 March 2020, NAV experienced a 250 per cent increase in citizen inquiries as hundreds of thousands of workers were laid off or furloughed and needed information about unemployment benefits, sickness benefits, and other welfare entitlements. NAV's contact centre, which normally operated with approximately 850 support representatives across 15 locations processing around 15,000 calls daily, was overwhelmed despite the rapid hiring of 70 temporary staff. Frida absorbed a substantial portion of this surge: within a few weeks, the virtual agent answered over 270,000 inquiries from citizens concerned about their situation as it related to the pandemic. During peak traffic periods, Frida was handling incoming inquiries equivalent to the workload of approximately 220 full-time human advisors.

The response to Frida from the public during the pandemic was reported as overwhelmingly positive, with approximately 80 per cent of interactions successfully resolved without the need to escalate to a human service representative. This resolution rate was achieved despite the rapidly changing policy landscape, as the Norwegian government introduced emergency benefit extensions, simplified application procedures, and new temporary schemes in quick succession.

The AI Trainer team responsible for maintaining Frida's knowledge base played a critical role in the system's pandemic performance. The team comprised six non-technical NAV employees whose knowledge and understanding of the agency's policies and procedures made them uniquely qualified to keep Frida operating at maximum capacity. During the crisis, the team updated Frida's responses multiple times daily as government policies evolved, coordinated with NAV's website team to ensure consistency between chatbot responses and official web content, and ensured that messaging maintained an empathetic and contextually appropriate tone given the distress many citizens were experiencing. The entire maintenance operation was conducted remotely with zero downtime.

NAV's Contact Centre Director publicly stated that the agency could not have managed the pandemic surge without Frida, highlighting the system's role as an essential component of the agency's crisis response infrastructure rather than merely a convenience feature.

Beyond the pandemic, Frida has continued to operate as a permanent component of NAV's multi-channel service delivery architecture. The chatbot serves as a first point of contact within NAV's digital services, functioning alongside the nav.no website, the "My Page" (Min side) self-service portal, telephone services, and in-person NAV offices. Academic research conducted by the University of Agder in collaboration with NAV has examined the human-AI interaction dynamics of the Frida system, particularly the handover process between the chatbot and human advisors, identifying opportunities for improving the transition experience and leveraging the information gathered by Frida during initial interactions to improve the efficiency of subsequent human-assisted service delivery.

The system's operational model — combining automated first-line service with structured human escalation — has been recognised internationally as an example of how conversational AI can support government service delivery, particularly in crisis situations. The OECD has cited Norway's use of AI in public services, and Frida has been referenced in Nordic comparative studies of public sector conversational AI adoption alongside similar implementations in other Scandinavian welfare agencies.

Frida represents a notable case of conversational AI deployment in social protection because of the breadth of welfare programmes it covers, the scale of the population it serves (Norway's 5.5 million residents who may interact with NAV at various life stages), and the demonstrated resilience of the system under extreme demand conditions during the pandemic. The system's governance model — with non-technical domain experts maintaining the AI's knowledge base rather than relying solely on technical AI engineers — offers a distinctive approach to public sector AI management that prioritises policy accuracy and citizen-appropriate communication over technical sophistication.

Classifications follow the DCI AI Hub Taxonomy. Hover over field labels for definitions.

Social Protection Functions

Implementation/delivery chain
Outreach/communications/sensitisation primary
SP Pillar (Primary) The social protection branch: social assistance, social insurance, or labour market programmes. Social insurance
SP Pillar (Secondary) The social protection branch: social assistance, social insurance, or labour market programmes. Social assistance
Programme Name NAV Multi-Channel Service Delivery — Frida Virtual Agent
Programme Type The type of social protection programme, classified under social assistance, social insurance, or labour market programmes. View in glossary Other
System Level Where in the social protection system the AI is applied: policy level, programme design, or implementation/delivery chain. View in glossary Implementation/delivery chain
Programme Description NAV (Norwegian Labour and Welfare Administration) deploys the Frida conversational AI chatbot as part of its multi-channel citizen service architecture, providing 24/7 automated assistance on social insurance and social assistance benefits including unemployment, sickness, pensions, child support, and disability. Built on Boost.ai's platform, Frida handles first-line inquiries with escalation to human advisors during business hours.
Implementation Type How the AI output is produced: Classical ML, Deep learning, Foundation model, or Hybrid. Affects validation, compute requirements, and governance profile. View in glossary Classical ML
Lifecycle Stage Current stage in the AI lifecycle, from problem identification through to monitoring, maintenance and decommissioning. View in glossary Monitoring, Maintenance and Decommissioning
Model Provenance Origin of the AI model: developed in-house, adapted from open-source, commercial/proprietary, or accessed via third-party API. View in glossary Commercial/proprietary
Compute Environment Where the AI system runs: on-premise, government cloud, commercial cloud, or edge/device. View in glossary Commercial cloud
Compute Provider The specific cloud or infrastructure provider hosting the AI system. Boost.ai
Sovereignty Quadrant Classification of data and compute sovereignty: I (Sovereign), II (Federated/Hybrid), III (Cloud with safeguards), or IV (Shared Innovation Zone). View in glossary III — Compute-Intensive Cloud with safeguards
Data Residency Where the data used by the AI system is stored: domestic, regional, or international. View in glossary Not documented
Cross-Border Transfer Whether data crosses national borders, and if so, whether documented safeguards are in place. View in glossary Not documented
Decision Criticality The rights impact of the decision the AI supports. High criticality requires HITL oversight; moderate requires HOTL; low may operate HOOTL. View in glossary Low
Human Oversight Type Level of human involvement: Human-in-the-Loop (active review), Human-on-the-Loop (monitoring), or Human-out-of-the-Loop (periodic audit). View in glossary HOTL
Development Process Whether the AI system was developed fully in-house, through a mix of in-house and third-party, or fully by an external provider. View in glossary Mix of in-house and third-party
Highest Risk Category The most significant structural risk source identified: data, model, operational, governance, or market/sovereignty risks. View in glossary Operational and system integration risks
Risk Assessment Status Whether a formal risk assessment, informal assessment, or independent audit has been conducted for this system. Not assessed

Risk Dimensions

Operational and system integration risks

Impact Dimensions

Autonomy, human dignity and due process
Equality, non-discrimination, fairness and inclusion
Systemic and societal
  • Human oversight protocol
CategorySensitivityCross-System LinkageAvailabilityKey Constraints
Unstructured and text-based contentNon-personalSingle source (no linkage)Currently available and usedFrida's knowledge base is curated from NAV's official policy documentation, website content, and procedural guidance. Requires continuous manual updating by the AI Trainer team to reflect policy changes, particularly during rapidly evolving situations such as the COVID-19 pandemic

Følstad, A. and Taylor, C. (2022) 'Developing human/AI interactions for chat-based customer services: lessons learned from the Norwegian government', European Journal of Information Systems, 32(4), pp. 601–620. doi: 10.1080/0960085X.2022.2096490.

View source Academic journal article

Boost.ai (2024) 'Conversational AI supports Norway's COVID response', Boost.ai Case Studies. Available at: https://boost.ai/case-studies/how-conversational-ai-is-helping-norways-citizens-with-covid/ (Accessed: 26 March 2026).

View source Other

Boost.ai (2024) 'Nordic Public Sector Conversational AI Case Study', Boost.ai Case Studies. Available at: https://boost.ai/case-studies/public-sector-nordics-conversational-ai-case-study/ (Accessed: 26 March 2026).

View source Other
Deployment Status How far the system has progressed into real-world operational use, from concept/exploration through to scaled and institutionalised. View in glossary Full Production Deployment
Year Initiated The year the AI system was first initiated or development began. 2018
Scale / Coverage The scale and geographic or population coverage of the deployment. National — available to all citizens via nav.no. During COVID-19, handled over 270,000 inquiries in weeks, equivalent to workload of 220 full-time advisors. NAV serves Norway's 5.5 million residents and normally processes approximately 15,000 contact centre calls daily across 850 representatives at 15 locations.
Funding Source The source(s) of funding for the AI system development and deployment. Unknown
Technical Partners External technology vendors, academic partners, or development partners involved. Boost.ai (Norwegian conversational AI platform provider) supplies the underlying virtual agent technology. NAV maintains a six-person non-technical AI Trainer team for knowledge base management. University of Agder has conducted collaborative research on the system's human-AI interaction dynamics. (Boost.ai; University of Agder)
Outcomes / Results During COVID-19, Frida answered over 270,000 citizen inquiries within weeks, with approximately 80% of interactions resolved without escalation to human advisors. At peak traffic, handled workload equivalent to approximately 220 full-time employees. NAV's Contact Centre Director stated the agency could not have managed the pandemic surge without Frida. (Boost.ai case study)

How to Cite

DCI AI Hub (2026). 'NAV "Frida" AI Chatbot for Social Benefits', AI Hub AI Tracker, case NOR-001. Digital Convergence Initiative. Available at: https://socialprotectionai.org/use-case/NOR-001 [Accessed: 1 April 2026].

Change History

Created 30 Mar 2026, 08:41
by v2-import (import)