What is this course about?
A citizen who cannot get an answer from an agency cannot go to a competitor, and that asymmetry changes every design decision. This course covers plain language and reading level as requirements, Section 508 conformance for conversational interfaces, language access, and escalation to a human treated as a right rather than a fallback.
Who is this course for?
- Digital service and citizen experience teams
- Program staff running public-facing services
- Accessibility coordinators
- Agency communications staff
What do I need before starting?
- Responsibility for a public-facing service
- No technical background required
What will I be able to do afterwards?
- Explain why the absence of an alternative provider changes design obligations
- Write agent responses at an appropriate reading level
- Meet Section 508 and WCAG conformance in a conversational interface
- Design escalation that a constituent can always reach
- Publish accuracy and performance metrics honestly
What does each module cover?
Why does the absence of a competitor change everything?
40 minThe obligation that follows from being the only provider of a service someone needs.
Objectives
- Articulate the obligations created by monopoly provision
- Identify the design choices this rules out
- Set the service standard before building
Topics
Activity. List the design choices a commercial service could make that your agency cannot.
What reading level should an agency agent write at?
45 minPlain language obligations and the measurement that verifies compliance.
Objectives
- Apply plain language requirements to generated text
- Measure and control reading level
- Keep precision while simplifying
Topics
Activity. Measure your agent's output reading level and bring it into range without losing accuracy.
How do you meet Section 508 in a chat interface?
55 minConformance for streaming, dynamically updating conversational output.
Objectives
- Apply WCAG to a streaming chat interface
- Manage focus and screen reader announcements
- Produce conformance evidence
Topics
Activity. Test your interface with a screen reader and log every conformance failure.
How do you meet language access obligations?
45 minServing constituents in their own language at the quality the obligation requires.
Objectives
- Determine language access obligations for your service
- Assess translation quality per language
- Route high-stakes content to human translation
Topics
Activity. Assess translation quality for your service's top three languages.
How does a constituent always reach a human?
50 minEscalation designed so nobody is trapped in a loop with an agent that cannot help.
Objectives
- Make human escalation reachable at any point
- Detect the constituent who is stuck
- Preserve context through handoff
Topics
Activity. Attempt to get trapped in your own agent and fix every path where you succeed.
How do you handle eligibility and benefits questions?
50 minThe highest-stakes category, where a wrong answer costs a constituent money or access.
Objectives
- Identify eligibility questions requiring human determination
- Ground responses in current program rules
- Design disclaimers that are honest and useful
Topics
Activity. Classify thirty real constituent questions by whether an agent may answer them.
What should you publish about performance?
40 minPublic reporting on accuracy and escalation as an accountability practice.
Objectives
- Select metrics appropriate for public reporting
- Report honestly including failures
- Design the public transparency page
Topics
Activity. Draft the public performance page for your service.
Auditing a citizen-facing agent
55 minThe workshop module: a full accessibility and escalation audit with remediation.
Objectives
- Run a complete accessibility and escalation audit
- Prioritize and remediate findings
- Establish an ongoing audit cadence
Topics
Activity. Complete the audit with an assistive technology user and remediate the top findings.
What is the capstone project?
Citizen service agent audit and remediation package
Audit a public-facing agent for reading level, Section 508 conformance, language access, escalation reachability, and eligibility boundaries — then remediate the findings and draft the public performance reporting page.
Deliverable: An audit report with remediations applied and a published performance page.
How are learners assessed?
- Accessibility audit conducted with a real assistive technology user
- Escalation trap test — no path may leave a constituent stuck
- Eligibility classification reviewed by program staff
What ships with the course?
Facilitator guide
Session-by-session running order, discussion prompts, and the questions that reliably derail a room.
Learner workbook
Exercises, checklists, and the templates each module's activity produces.
Hands-on lab environment
A sandboxed ibl.ai deployment so exercises run against real agents, not screenshots.
Assessment bank
Scenario questions and rubric criteria mapped to each stated learning outcome.
Source bibliography
Every primary regulation and standard cited on this page, linked and dated.
Which AI agents does this course use?
The hands-on modules run against agents already deployable on the ibl.ai platform for government.
Where does the course material come from?
Every module is grounded in primary sources — the regulation, standard, or research itself, not a summary of it. Each was resolved at authoring time.
- Section508.gov
GSA
Federal accessibility requirements and conformance reporting.
- Web Content Accessibility Guidelines
W3C
Technical conformance target for the chat interface.
- AI Guide for Government
GSA Centers of Excellence
Federal practice guidance for citizen-facing AI.
- Artificial Intelligence
CISA
Security requirements for public-facing agency AI.
Delivery notes
Binding guidance for anyone preparing and delivering this course.
- Module 3 requires a real assistive technology user. A sighted facilitator tabbing through an interface produces a false pass, and streaming chat has failure modes that only a screen reader user encounters.
- Module 5's trap test is the course's best exercise. Facilitators should try genuinely hard to get stuck, because constituents will.
- Module 6 must be conservative. Eligibility determinations have legal consequences for constituents and the default should be human determination.
- Module 7's public reporting will meet institutional resistance. Frame it as risk reduction — an agency that publishes its error rate is harder to embarrass with it.
- Use a real agency service throughout. Generic examples let participants avoid confronting their own service's failures.
Why run AI training on a platform you own?
You own the course, not a licence to it
Course content, learner data, and the platform run inside your perimeter — you own all the code and the data.
Model-agnostic delivery
Run the course's AI components on any LLM — Claude, GPT, Llama, Gemini, Command — and switch anytime.
No per-seat training licences
Usage-based or self-hosted, so cost tracks actual use rather than headcount.
Deploy anywhere
Cloud, private VPC, on-premise, or fully air-gapped — including for cohorts that cannot use public AI tools.
Frequently asked questions
What does the Citizen Service Agents: Design, Escalation, and Accessibility course cover?
A citizen who cannot get an answer from an agency cannot go to a competitor, and that asymmetry changes every design decision. This course covers plain language and reading level as requirements, Section 508 conformance for conversational interfaces, language access, and escalation to a human treated as a right rather than a fallback. It runs 6 hours across 8 modules across 8 modules, at intermediate level, and closes with a capstone: Citizen service agent audit and remediation package.
Who should take Citizen Service Agents: Design, Escalation, and Accessibility?
It is written for Digital service and citizen experience teams, Program staff running public-facing services, Accessibility coordinators, Agency communications staff. Prerequisites: Responsibility for a public-facing service; No technical background required.
Can we run this course on our own infrastructure?
Yes. ibl.ai is model-agnostic and deploy-anywhere — cloud, private VPC, on-premise, or fully air-gapped — and you own all the code and the data. Cohort data, submissions, and any material learners upload stay inside your perimeter, which matters for government teams that cannot send work to a public AI tool.
How do we get access to Citizen Service Agents: Design, Escalation, and Accessibility?
Request access and we will set it up for your cohort — hosted by ibl.ai, or running against your own deployment. Tell us the group size and timing you need, and whether it should run inside your own perimeter.
How much does AI training for government cost on ibl.ai?
There is no per-seat pricing — you pay for usage or self-host and pay only for the infrastructure, so a 5,000-person rollout does not cost 5,000 licences. 1.6M+ users across 400+ organizations run the platform this way, including NVIDIA, MIT, and Syracuse University.