From Long Lines to Smart Triage: Piloting AI-Powered Client Screening for Delaware DSS.

About the Client

DSS, within Delaware’s Department of Health and Social Services, operated 18 local benefits offices across the state, each staffed at the service counter by senior social workers — the only staff who fully understood the state’s benefit-related rules and could run the software used to calculate benefit levels. This engagement built directly on Iknow’s earlier client survey work for DSS, which had already identified long wait times as clients’ primary front-office concern.

Industry Context

Rules-based and AI-driven eligibility screening has since become a mainstream tool across the social services sector. In 2025, Los Angeles County’s Imagine LA, working with Nava Labs, piloted a generative-AI-powered assistive chatbot to help caseworkers navigate public benefits eligibility, part of a broader effort to close an estimated $227 billion in unclaimed public benefits nationally each year. At the same time, the U.S. Social Security Administration rolled out generative-AI-powered chatbots for both the public and its own staff as part of a “digital-first” customer service strategy. A 2026 framework from the American Public Human Services Association explicitly endorses the design philosophy behind Delaware’s 2016 pilot: applying “the right level of technology to the right use case,” with rules-based logic handling well-defined, high-volume processes and more advanced AI layered in only where it adds real value. The goal is technology that assists staff rather than replacing their judgment. Delaware’s Neota Logic-based triage system, and the statewide rollout it earned through federal grant funding, anticipated that now-standard approach by nearly a decade.

Current Situation

DSS’s Division Director recognized that staffing senior social workers at every service counter was both inefficient and costly, since these staff were needed for complex casework but were instead spending significant time on routine walk-in questions. He asked Iknow to propose a solution that could better allocate that expensive expertise without compromising the accuracy or consistency of service clients received.

• Undifferentiated queuing. Every walk-in client waited in the same line regardless of whether their need was simple (dropping off an application, picking up a form) or complex.
• Expensive, essential expertise tied up at the front counter. Senior social workers were DSS’s most costly front-line staff, yet they were the only ones who fully understood the state’s complex benefit rules and could run the benefits determination software.
• No mechanism to distinguish need at arrival. DSS had no way to identify quick-response cases at the point clients walked in, so even simple requests waited alongside complex ones.
• Inconsistent service statewide. The accuracy and completeness of guidance clients received varied with individual staff training, skill, and experience across DSS’s 18 offices.
• Deep, specialized knowledge to capture. Any solution had to reliably encode the full complexity of SNAP eligibility rules and DSS policy well enough for less expensive front-desk staff to operate.

• Reduce the number of walk-in clients who had to wait in a queue for service.
• Free senior social workers to spend more time on actual benefits casework rather than front-counter triage.
• Deliver more consistent, accurate guidance to clients regardless of office or staff member.
• Validate the approach through a controlled pilot before committing to a statewide investment.
• Support DSS in securing funding for a statewide rollout if the pilot proved successful.

How Iknow Structured the Work

Iknow moved from direct observation of real office operations, through knowledge engineering and platform selection, to an agile rule-development process validated by a controlled, two-office pilot — reserving any recommendation for statewide investment until the approach had proven itself in practice.

Key Activities & Decisions

• Observation & Opportunity Identification. Iknow observed walk-in traffic patterns at several local offices and identified that triaging clients immediately upon arrival — handling “quick response” needs like directing clients where to drop off applications or pick up forms — could significantly reduce the number of clients who had to wait at all.
• Knowledge Engineering. Iknow interviewed senior social workers familiar with front office operations and concluded that the knowledge needed for triage could be captured as a defined set of decision rules. Those rules drew on DSS’s policy manuals, training materials, standard operating procedures, and program forms and handouts, and incorporated both U.S. Food and Nutrition Service and Delaware-specific SNAP eligibility rules and procedures.
• Platform Selection. Iknow evaluated open source and commercial rules engine software and selected Neota Logic’s reasoning engine platform for its rules-based functionality, complex reasoning, and process workflow capabilities — including support for initial eligibility screening, dynamically optimized, minimal-question triage sequencing, and personalized next-step guidance for each client.
• Agile Rule Development. Iknow built and refined the decision rules through a series of workshops with DSS experts, using an agile approach to quickly incorporate their feedback.
• Pilot Testing. Once a sufficient set of rules was developed, Iknow coded them into the Neota Logic platform and piloted the system on laptops connected to the hosted platform over local office WiFi, testing at two local offices until the triage system demonstrated sufficient accuracy.
• Grant Support. Following the pilot’s success, Iknow worked with DSS’s Grant Office to prepare the business case and detailed statement of work supporting a statewide funding request.

Stakeholders & Collaboration

Iknow worked directly with DSS’s Division Director, senior social workers who served as the subject-matter experts behind the triage rules, and DSS’s Grant Office, while the completed pilot and grant application were ultimately reviewed and funded by the U.S. Department of Agriculture’s Food and Nutrition Service.

Capturing Deep Expertise Reliably Enough to Delegate It

The knowledge needed for triage existed only in the heads of DSS’s most experienced senior social workers, making it difficult to delegate to less expensive front-desk staff without losing accuracy. Iknow addressed this through systematic knowledge engineering — grounding the rules in DSS’s own policy manuals, training materials, and procedures, and refining them iteratively through direct workshops with the experts themselves — before coding anything into the platform.

Eliminating Statewide Inconsistency

Because individual staff training, skill, and experience varied across DSS’s 18 offices, clients had historically received inconsistent guidance depending on which office or worker they encountered. Iknow addressed this by encoding the triage logic directly into a rules-engine platform rather than relying on informal staff training, so every client received the same accurate, complete guidance no matter which office they visited.

Proving the Model Before Committing to Statewide Investment

A new, AI-powered front-office operating model carried real risk if rolled out statewide before it was proven. Iknow addressed this by running a controlled pilot at two representative offices, testing the system until it demonstrated sufficient accuracy, and only then supporting DSS’s request for the funding needed to expand it further.

Operational Outcomes

• About 60 percent of walk-in clients no longer had to wait in line at all, and clients who did need a senior social worker experienced significantly shorter wait times.
• Triage interactions became more accurate, consistent, and inclusive of all relevant benefit programs, with no irrelevant or redundant questions, since the software presented each next question based on the client’s prior responses.
• The system generated a personalized list of required data and documentation for each potential benefits program, helping clients progress more easily to their next step.

Strategic and Organizational Outcomes

From DSS’s perspective, the triage logic was accurate, up to date, and fully consistent statewide — eliminating the variability that had previously existed due to differences in staff training and experience — while dynamically minimizing the number of questions asked and tracking common roadblocks for further refinement. DSS representatives appeared more knowledgeable and credible to clients, and DSS’s costs fell because fewer expensive senior social workers were needed at the front desk; those redeployed to one-on-one casework became more efficient by handling only complex cases. Based on the pilot’s success, Iknow helped DSS’s Grant Office prepare the business case and statement of work for a statewide rollout, and in August 2016 the USDA’s Food and Nutrition Service awarded Delaware DSS a Process and Technology Improvement Grant of $667,618 to expand the triage system to all 18 local offices.

Timeline to Impact

Iknow completed knowledge engineering, platform selection, rule development, and pilot testing, and supported a successful statewide grant application, within the nine-month engagement — moving from an early-stage concept to a funded, statewide expansion plan within a single project.

Core Skills

• AI & Knowledge Engineering
• Business Process Innovation
• Government & Public Sector Consulting
• Grant & Funding Support

Methods & Frameworks

• Knowledge engineering and decision-rule capture
• Agile rule development and expert workshops
• Controlled pilot testing and validation
• Grant business case and statement of work development

Technologies & Tools

• Neota Logic reasoning engine platform (rules-based logic, complex reasoning, process workflow)