From the Campus Operations edition of November 10, 2026
DePaul's staff-facing financial aid AI agent cuts counselor prep from about 15 minutes to about one, per an EDUCAUSE Review case study
DePaul University with Salesforce (EDUCAUSE Review case study, Mission Partner collaboration) · Higher education administrative operations (financial aid office) · United States (Illinois)
- Publisher
- EDUCAUSE Review
- Original publication
- 10 November 2026 (EDUCAUSE Review)
- Source retrieved
- 2026-11-11
What happened
DePaul built a staff-facing financial aid agent on Salesforce Agentforce. In the first minute of an advising appointment it produces a financial-aid overview drawn from four separate sources (DePaul knowledge articles, federal guidance, Illinois guidance, and student data in Data360) plus an action plan. The case study says counselor administrative burden fell from fifteen minutes to about one minute, that processing time for financial aid tasks fell by an average of 90 percent, and that over one hundred hours were freed since launch in November 2025. Build took eight weeks over three iterations after a four-week discovery that interviewed every counselor and identified more than sixty counselor actions.
Why it matters
A concrete administrative-operations pattern with stated numbers: keep the agent staff-facing, ground it in existing governed data, and decide up front how freed time is reused. Fresh today (published 10 November 2026) and directly about back-office workload on campus.
Evidence and measured results
Self-reported by the institution through an interview with DePaul's VP for Information Sciences, published by EDUCAUSE in a Salesforce collaboration. No baseline method, sample size, accuracy or error-rate data, or audit are described; the 90 percent figure is an average with no stated definition of 'task'.
Limitations and uncertainty
Vendor-partnered case study with no independent verification; single institution; the financial aid office is student-adjacent, so relevance to wider back-office operations is partial; US federal and Illinois aid rules shape the design; no cost, error or staffing-impact figures.
Put this evidence to work
Lighthouse Advisory interpretation, grounded in this source. Enriched 2026-11-10; this does not change the original publication date. Each role follows the same checklist so takeaways compare across sources.
Sales
Role takeaway
- Customer problem
- Advisers spend appointment time assembling data from several systems and policy sources instead of resolving the student's problem.
- Stakeholders
- CIO or VP of information systems
- Director of financial aid or equivalent service office
- Registrar or student accounts lead
- Data governance owner
- Discovery questions
- How long does a counselor spend assembling context before advising today, and how was it measured?
- Which data sources must be combined and who owns each?
- How would freed time be reused and who decides?
- Do staff have a say in what the agent does?
- Value hypothesis
- A staff-facing assistant that prepares a case summary and suggested actions may shorten prep time and reduce repeat visits, if baselines are measured first.
- Potential engagement
- A bounded discovery and proof of value for one high-volume office: map staff actions and data sources, measure baseline prep time, and test an assistant on a sample.
- Unsupported claims
- Guaranteed time or cost savings
- Accuracy or error rates
- Headcount effects
- That the 90 percent figure generalizes beyond DePaul financial aid
Pre-sales engineering
Role takeaway
- Fit
- Good fit where staff consult several disconnected sources per case and a governed data layer already exists; weaker without one.
- Architecture and integration
- Agent over knowledge articles, federal and state guidance, and student records, with an LLM chosen per task; DePaul reused an existing Data360 foundation.
- Prerequisites
- Governed, integrated student data layer
- Inventory of staff actions and authoritative policy sources
- Data decisions on scope, aggregation and latency before build
- Constraints
- Federal and state aid rules vary by institution and jurisdiction
- Reported platform is Salesforce Agentforce; other estates need separate evaluation
- Student records are regulated education records
- Security
- Keep the agent staff-facing as DePaul did; review role-based access, logging and retention, which the source does not describe.
- Proof of value
- Compare agent summaries with human-prepared summaries on a sample
- Measure appointment prep time before and after
- Red-team for unauthorized data exposure and wrong guidance
- Track repeat-visit rates
Delivery
Role takeaway
- Work
- Run staff interviews, define the agent's tasks, make data decisions, build in iterations, red-team, launch and track time reuse.
- Dependencies
- Data foundation
- Implementation partner or in-house skills
- Office leadership buy-in
- Policy source owners
- Ownership
- Office director owns the workflow and outcomes; IT owns the platform and data pipelines.
- Skills and adoption
- Counselors need to learn to verify and adjust AI action plans; new staff onboarding may change.
- Governance checkpoints
- Goal and time-reuse agreed before build
- Data decisions signed off
- Red-team results reviewed before launch
- Periodic accuracy review after launch
- Acceptance criteria
- Prep time reduced against a measured baseline
- Agent summary matches human reference on an agreed sample
- No unauthorized data exposure in red-team tests
- Repeat-visit rate tracked
- Risks
- Self-reported results may not reproduce
- Outdated policy sources produce wrong guidance
- Time saved is absorbed by higher volume rather than reinvested
- Staff distrust without involvement
Implementation considerations
Lighthouse Advisory interpretation across the operating dimensions a public-sector buyer must settle before this evidence becomes a design. Each note answers the question under its heading for this specific source.
Architecture and integration
What must connect, and where does the AI sit in the workflow?
Agentforce agent over four disconnected sources; an existing Data360 environment, built earlier for a predictive-analytics project, served as the data foundation. A third-party implementation partner selected an LLM per task and connected the data sources.
Governance
Who approves, reviews and stays accountable for outcomes?
The team states good data governance is a prerequisite, key data decisions should be made before starting, and the goal of increasing human connection should be discussed early with counselors. It also recommends extensive red-team testing, including checks on what the agent must not do.
Security and privacy
What data, permissions and controls need testing?
Staff-facing design is presented as limiting privacy and security risk: a student-facing agent could expose other students' data in a malfunction. The case study gives no detail on logging, retention or access controls.
Accessibility and workforce
Who is affected, and what skills or accommodations follow?
Counselors were interviewed during discovery and the time saved is framed as redirected to student contact, not headcount reduction. The article reports no staff survey or union input.
Procurement
What should contracts, pricing and exit terms secure?
Platform plus third-party implementation partner; the article gives no pricing or licensing terms.
Operating model
Which teams own the service once it runs?
Joint IT and financial aid leadership, a technical lead who was a long-serving financial aid data analyst, and a partner. Reuse of the freed time is named as a deliberate management decision.
Publication history
- 2026-11-10Campus Operations · Issue 622 resources
Stable resource ID: depaul-staff-facing-financial-aid-agent-educause-20261110