What happens between finding an idea useful in a course and trying it in a classroom? Kennedy and colleagues explore this question in The Reflect-And-Act Chatbot: Fostering Applied Professional Learning for Teachers in Challenging Environments. Their study suggests that a chatbot can prompt teachers to reflect on learning and identify possible applications.
It also exposes a difficulty: intentions often need more detail before they become workable plans. What interests me is the attention to this transition in professional learning. The version tested used scripted responses; a more adaptive, LLM-powered version was proposed as a future development.
What Did the Researchers Test With Rebo?
The researchers developed Rebo within online professional learning programmes called Co-designed Massive Open Online Collaborations, or CoMOOCs. These programmes emphasise teachers’ contributions, shared experience, and learning relevant to their circumstances.
Rebo was introduced in courses on blended and online learning design and education in conflict and crisis settings. The latter was co-designed with displaced educators on the Thai–Myanmar border. Rebo was offered in English and Burmese in that course.
Using a design-based research approach, the team examined chatbot conversations and course forum comments. Rebo asked participants what they had found useful and where they could apply it. It also offered an optional reminder and a later conversation about what happened.
Participants generally described the chatbot as accessible and helpful for articulating ideas, although some reported technical problems. Analysis of their plans found limited detail about implementation, measures of progress, constraints, and timing.
The evidence supports Rebo’s feasibility as a reflection and planning tool. Evidence of application was much thinner: relatively few follow-up conversations were available, and these provided self-reported accounts.
What Can Teacher Educators Learn From This Study?
For those working in AI and teaching practice, this paper offers a useful way to think about the design of professional learning support. Its strongest insights concern the questions teachers need help answering as they prepare to use course ideas.

Build Application Into the Learning Experience
Rebo’s initial conversation connected course content to a particular teaching situation. Participants were invited to identify something useful and consider where they could use it.
I find that connection worth building deliberately into professional development. A teacher can understand a strategy and still need time to work out where it fits. The paper gives that planning work a visible place in the course. It also creates an opportunity for course designers to learn which ideas participants see as relevant to their work.
Help Teachers Add the Missing Details
The gap between intention and planning is perhaps the most useful finding. Participants could name activities they wanted to try, but their plans often lacked details that would make implementation and review easier.
The researchers used an adapted SMART framework to examine whether plans were specific, measurable, activity-related, realistic, and time-bound.
My takeaway is that reflection prompts need room for follow-up questions: What would the teacher actually do? What resources or constraints matter? When could the activity happen? What evidence would help them judge the result? These questions are implications of the planning gaps identified in the study; the tested chatbot did not yet provide adaptive support to resolve them.
Return to the Plan After Teachers Have Tried It
Rebo’s proposed cycle included a return conversation about how implementation went and what participants would take forward. The limited follow-up accounts offered glimpses of teachers adapting their practice, including seeking learners’ feedback on online learning.
This stage deserves as much design attention as the initial reflection. The study leaves open how to encourage teachers to return, and some participants reported missing reminder emails. A follow-up mechanism needs to work reliably before researchers can assess whether it helps teachers sustain a cycle of planning, action, and adjustment.
Keep Teachers’ Knowledge of Their Context Central
The courses were grounded in the idea that professionals understand the settings in which they work. That principle is particularly significant in conflict, displacement, and resource-constrained environments.
The authors propose using a future LLM-enhanced Rebo to help teachers clarify their goals while retaining responsibility for decisions. I would keep that boundary central to its evaluation. A more detailed plan still needs to fit the teacher’s learners, available resources, and circumstances.
Test the Educational Purpose Before Expanding the Technology
The team first tested a simple version to investigate whether structured conversation was useful. Their proposed LLM integration responds to identified limitations, including static replies and insufficient support for goal development.
I appreciate this sequence. It gives future development a concrete educational purpose. The next question is whether more adaptive dialogue helps teachers produce workable plans and return to examine what happened when they used them.
References
Kennedy, E., Guo, X., Kopkow, A., Littlejohn, A., & Pammer-Schindler, V. (2026). The Reflect-And-Act Chatbot: Fostering applied professional learning for teachers in challenging environments. Innovation and Education, 7(1), 58–85. https://doi.org/10.1163/25248502-bja00010




