A school website chatbot can help visitors find published admissions requirements, enrollment dates and office contacts. Start with a small set of public pages and check its answers before making it available to families. Agentkit's free plan includes one chatbot and 50 messages per month. That allowance is not 50 complete conversations. Check the published plan limits.
This guide focuses on a public-information pilot. It does not establish that Agentkit is suitable for student records or approved for your institution. Agentkit publishes this guide and sells the product described here; no school deployment or admissions outcome was tested for this article.
Why schools are adding chatbots to their websites
Public admissions pages are a useful starting point because the office can compare each answer with an approved source. A school can choose a narrow question set before deciding whether to expand.
There are real university examples of this use case. Penn State describes LionChat answering enrollment questions on its admissions, student aid, registrar and bursar sites, with an inquiry form when it cannot answer. This is Penn State's deployment, not an Agentkit customer example. Penn State's announcement.
School website chatbot use cases
| Pilot topic | Example question to test | Boundary to check |
|---|---|---|
| Admissions | Where is the published application deadline? | Correct program and intake; link to the current source |
| Enrollment | Where can I find the required application documents? | Explain the public list; do not decide an applicant's eligibility |
| Fees | Where is the published tuition schedule? | Link to the schedule; route individual balances to the school's secure service |
| Office contacts | How do I contact admissions? | Give the approved contact route, including out-of-hours expectations |
| Questions in another language | Can you explain the published deadline in Spanish? | Ask a fluent reviewer to check meaning and dates before launch |
These are proposed test cases, not observed Agentkit results. Record the answer you receive and the source used to check it.
What to train your chatbot on
Source selection affects answer quality. Start with current, approved public information and check what was actually extracted.
| Source type | What to include | Why it matters |
|---|---|---|
| Website crawl | Admissions pages, academic departments, financial aid, student services, campus life | Covers the broadest range of questions visitors actually ask |
| Documents | Approved public course catalog or enrollment guide; check supported file types before uploading | Contains detailed structured information that web pages often summarize |
| Q&A pairs | Staff-written questions and approved answers, with personal details removed | Fills gaps where published content does not directly answer common questions |
| Plain text snippets | Office hours, key dates, contact information, policy summaries | Quick-reference information students ask about constantly |
Ask admissions staff which questions deserve a place in the pilot. Write approved answers, remove personal details and compare the chatbot's answers with that reference set. Adding a source does not by itself prove the answers are accurate.
Follow the website-source instructions to add and check the selected pages.
Prepare the source and answer review before setup
Use the school source register CSV to assign an owner and approval status to each public page. Then fill in the admissions answer-test CSV. Both files open in a spreadsheet. They are blank working templates with suggested checks, not results from a school deployment.
For each source, record the program and intake it applies to. Copy the exact current fact into the test's expected-answer field and include its public source URL. An application deadline without the intake or program is not enough to mark an answer correct. Ask the content owner to approve the reference answer before testing the chatbot.
Run the same question in the Playground and on the approved embedded test page. Record the actual answer, date and reviewer verdict separately for each run. A passing result needs the right fact, the right scope and an appropriate source or contact route. Leave tests not yet run marked untested. Recheck affected rows after changing a source; a prior pass does not cover the new content.
When the reference answers are ready, create a free chatbot for the pilot and follow the setup steps below within the published allowance. Keep student information out of both spreadsheets and test prompts.
Setting up your school website chatbot
Use this sequence to plan and review a pilot. The steps below follow public Agentkit documentation; they have not been run on a school site for this article.
- Create an Agentkit chatbot and choose a small public question set. Confirm that your school approves the trial before sharing the widget with families.
- Add the specific public pages that answer those questions. The free plan permits up to 10 website links. Check that each important source was added and contains the needed text. Website sources.
- Test the questions in the Playground. Compare dates, program names and contacts with the source pages. Include a question the sources cannot answer and check the response rather than assuming it will reach a person.
- Copy the embed snippet from the agent's Channels page for a school-approved test page. Use the generated snippet rather than a hardcoded CDN example. Widget embed documentation.
- Test the embedded version as well. Confirm that the widget loads, the selected sources answer correctly and the school's contact route remains available when the chatbot cannot help or reaches its allowance.
A small admissions pilot checklist
| Check | Record before launch |
|---|---|
| Current deadline | Question, source URL, intake, answer and reviewer verdict |
| Missing information | An unanswered question and whether the reply gives the approved office contact |
| Changed deadline | Update the source, retrain, and record whether the revised date appears |
| Language | A fluent review of the same question in a language families use |
| Allowance | A named owner for usage monitoring and an accessible alternative contact route |
This checklist is an illustrative review procedure. It contains no completed tests, pass rate or performance benchmark. Keep real student information out of test prompts.
Multilingual support for diverse student and parent populations
Include the languages used by your school's families in the pilot. Ask a fluent reviewer to check dates, eligibility wording and office names against the approved source. Do not treat a fluent response as proof that its meaning is correct, or assume that changing the source language preserves every detail. Keep a route to staff for questions that need interpretation or an individual decision.
Review privacy before a public pilot
Public source material does not establish that a chatbot deployment meets a school's privacy obligations. Visitors can type personal information, so the review must consider conversations as well as uploaded sources. This article does not establish Agentkit's FERPA or COPPA suitability.
Ask the school's responsible privacy and procurement staff to review the intended audience, contract, data collected, retention, access and deletion arrangements before launch. For a public-information pilot, use approved public pages and direct individual record questions to the school's established secure service. Do not invite student IDs, grades or account details in a public chat.
The U.S. Department of Education publishes guidance on privacy and online educational services. Its stated scope matters; it is not an approval of this product or a universal rule for every chatbot use.
Scaling across departments
Consider another department only after reviewing the pilot's answers and workload. Possible public-information topics include:
- Registrar: Registration procedures, transcript requests, graduation requirements
- Financial Aid: Aid policies, FAFSA guides, scholarship information
- IT Help Desk: Published IT help instructions and the approved support contact
- Housing and Residence Life: Application procedures, meal plans, move-in guides
- Career Services: Resume guides, interview tips, job fair schedules
- Library: Database access, interlibrary loan, study room reservations
Choose the scope and number of chatbots around content ownership and your plan limits. Department expansion needs its own source and answer review.
Plans and pricing for schools
Agentkit lists the following monthly plans as checked September 6, 2026:
| Plan | USD per month | Messages per month | Chatbots |
|---|---|---|---|
| Free | $0 | 50 | 1 |
| Hobby | $29.99 | 2,000 | 1 |
| Standard | $119.99 | 12,000 | 2 |
| Pro | $399.99 | 40,000 | 3 |
These are published allowances, not a forecast of how many families a school can serve. Check Agentkit billing documentation before choosing a plan.
For a broader budget, see the chatbot pricing guide.
Keeping content current
Assign an owner to each deadline, fee schedule and contact page. After a change, check that the source was retrained and test the affected answer. Agentkit documents daily website retraining on Standard and Pro; a scheduled refresh does not mean an urgent change is available immediately. Manual retraining is available for a specific updated page. Retraining instructions.
Measuring impact
Track answer correctness against the approved question set, questions that require staff help and the time spent maintaining sources. Conversation counts measure activity; they do not show that an inquiry was resolved or that staff time was saved. If you compare staff workload before and after a pilot, record changes in enrollment volume and office processes too. Report those observations without attributing every change to the chatbot.
For each review period, record the number of conversations checked, confirmed answers, requests for staff help and outcomes you could not verify. Keep unknown outcomes separate from confirmed resolutions. Use the chatbot KPI definitions and formulas to choose consistent denominators before comparing periods.
Frequently Asked Questions
Is there a free chatbot for school websites?
Yes. Agentkit lists a free plan with one chatbot and 50 messages per month. Use it to test a small source set, and check the current limits before planning a public service.
What can a school website chatbot do?
A public-information chatbot can help visitors find published admissions information and office contacts. Check its answers against approved sources and provide a clear contact route for individual cases. This guide's pilot checklist is a proposed review process, not a tested school deployment.
Are chatbots safe for student data?
Suitability depends on the actual service, data and institutional review. Training on public pages alone does not establish privacy compliance. Review conversations, retention, access and contractual terms with the school's responsible staff before launch.
How long does it take to set up a chatbot for a school?
There is no verified school setup-time benchmark for this guide. Source approval, answer checks, website access and privacy review all affect the schedule. Start with a limited pilot and finish those checks before making a launch commitment.
Can the chatbot handle questions in multiple languages?
Test the languages your families use and ask a fluent reviewer to check the result against approved information. This guide does not establish accuracy across a particular number of languages.
Do I need a developer to add a chatbot to my school website?
The widget uses an embed snippet generated from the agent's Channels page. The person managing your school website needs permission to add it. Follow the current embed documentation and test the result on an approved page before wider release.
Getting started
Start with a small set of public admissions pages, a named content owner and a written question set. Review the answers and the school's approval requirements before putting the widget in front of families.
Create a free Agentkit chatbot to evaluate the published free-plan allowance.
