Work stays in this tab. Save on this device or export before closing. Shared-device users may see saved work. Local drafts are not sent to a model or instructor.
Save explicitly to continue on another device. Account copies are private to you and authorized IHA staff; they are not public, sent to an AI model, or submitted for assessment. Use fictional or permitted, non-sensitive information.
Account saving is optional. Your browser work stays available without signing in.
Work here without an account. Save or export when you want to keep it.
Start with the people you serve
LESSON 01 / 06 ACTIVITY 01 / 08
Bring a team, nonprofit, community organization, or public-service mission. Learn to choose a useful problem, involve the people affected, and test AI with clear responsibility and evidence.
A mission growth plan: community need, outreach and partnership work, funding evidence, a bounded AI workflow, team capacity, and a review decision.
No prior AI experience is required. You need to read a short scenario and record or explain a response; a facilitator or peer can support reading and writing. Use a phone, computer, or the printable lesson pack. No purchase or AI account is required for the core course. Optional account saving and AI coaching have separate sign-in and usage requirements.
165–200 minutes of estimated lab practice across six lessons. These are planning estimates, not measured learner averages. Allow additional time for reading, decision checks, revisions, and optional fieldwork. Try one lesson per session; pause and resume at any activity.
Connect a specific service problem to your mission and the experiences of affected people.
Lab estimate: 25–30 minutes
Identify a manageable improvement with clear inputs, ownership, and approval boundaries.
Lab estimate: 25–35 minutes
Measure service quality and effort without overstating impact or collecting unnecessary information.
Lab estimate: 30–35 minutes
Evaluate a draft-only workflow against realistic failure cases before using it in a service.
Lab estimate: 30–35 minutes
Plan staff support, accessible participation, and a non-AI route through the service.
Lab estimate: 25–30 minutes
Make a documented decision that fits evidence, capacity, and community benefit.
Lab estimate: 30–35 minutes
The course tracks work recorded, not mastery. Authored answer feedback supports practice; written artifacts need human judgment. Self-review, reaching the last screen, or filling every field does not earn an IHA certificate or establish business results.
Use “Your work” to save on this device, export a backup, or optionally save to an account. On a shared device, consider an exported file instead. Return to this pathway and choose “Resume saved work”; account work loads through “Your work.” Device saves include your activity position. A paper learner can use the same outcomes and criteria.
Source-grounded clustering
Your AI workspace
Give it a task. Change one instruction. See what improves.
Run sends your instruction to the Academy AI provider and stores the run with your account. Sign-in and available usage required. Examples below are open to everyone.
Authored teaching examples, not live model runs or your results.
Before / First draft
Summarize what the community wants and propose a program.
Authored example output
The community wants more convenient evening workshops and better communications. An AI-powered scheduling and messaging system will meet these needs and reduce staff workload.
After / More useful draft
Group these notes into at most three themes. For each, list source IDs, differing needs, a small possible response, and one follow-up question. Keep staff efficiency separate from participant benefit. End by identifying a missing perspective. Do not claim consensus. Use the supplied source.
Authored example output
SCHEDULE ACCESS — P1/P2: evening and daytime preferences differ. Possible response: test interest in two times before choosing. Question: what specific days and barriers matter? READABLE INFORMATION — P3: small print is a barrier. Possible response: a large-text version and a phone route. Question: which format is usable? STAFF INFORMATION LOAD — S1: repeated calls create work. Possible response: a checked information sheet. Question: which questions repeat, and do callers find the current sheet? MISSING: people who never attend; the notes do not establish their reasons or community-wide agreement.
Take it further
Add a fictional non-attender who can only participate by phone. Update one theme and one proposed response without erasing the existing needs.
Add this variation to your instruction and run it again, or explain how you would revise the authored example. The starter prompt does not include this change automatically.
No live run? Evaluate the authored example and label your notes as example analysis.
Connect a specific service problem to your mission and the experiences of affected people.
Before using AI in your mission, distinguish the model, the app, and the workflow. A model generates outputs from learned patterns and the context it receives; an app may add search, files, or connections to other systems. A prompt requests behavior, but it does not itself grant permission or enforce access controls. A generated draft is not a verified source or a completed action. In this course you will define a small task, supply permitted information, inspect the result, and keep an authorized person responsible for release. The fictional exercises can be completed without buying or connecting a tool.
Begin with the change your organization exists to support. A tool purchase is an activity; making a service easier to access is a possible benefit. Name the people experiencing the problem and describe the moment it occurs. “We need AI” offers little direction. “People receive inconsistent instructions about how to attend our workshops” identifies something you can investigate.
Listen to participants and the people delivering the service. Ask about a recent experience: what they expected, what happened, what they tried, and what made the task difficult. Include people who use different languages, devices, or access methods. Staff experience matters, but it cannot stand in for every participant. Offer a way to contribute without sharing private circumstances or agreeing to use a new tool.
Separate an observation from your explanation. A missed appointment does not tell you whether the cause was an unclear message, transport, an inconvenient time, or a decision not to attend. Write down competing explanations and identify what would help distinguish them. AI can organize anonymized notes you are authorized to use; it cannot manufacture a community consultation or determine what the community wants.
Define a narrow problem statement with the people responsible for the service. State who is affected, what friction they experience, why it matters to the mission, and what remains unknown. A useful first response might be a clearer paper form or consistent source sheet. Choosing a simple improvement is a valid outcome of investigating AI.
Growth should mean more useful service or stronger capability, not simply a larger program. Investigate who is not being reached and why before adding sessions, buying tools, or asking for more money. Listening may reveal a schedule change, an accessible information route, a partner introduction, or a smaller service that better fits the mission. Use internal feedback only for the purpose people agreed to; a fundraising story requires separate permission.
Synthetic case: Harbor Community Center runs free workshops. In a practice scenario, a participant says the printed notice and reminder gave different start times. A staff member reports retyping details in three places. The team identifies inconsistent information as a hypothesis and plans to check source documents and invite feedback. It does not infer that all missed attendance was caused by reminders.
Watch for: Starting with a favored tool and treating a staff assumption as the community’s expressed need.
Help me organize these anonymized service observations. Separate reported experiences, my interpretations, and questions still unanswered. Do not invent participant views. Suggest one follow-up question and one possible non-AI improvement. Notes: [insert].
Optional AI support
Ask for a hint or feedback on your draft. AI feedback can be wrong. Check it against the source and review criteria; it does not award a grade or credential.
Sign-in and your existing AI allowance apply. What you send goes to the AI provider and is saved in your Sage Learning conversation history. Use fictional or permitted, non-sensitive material. Nothing is sent until you select the button below.