
Campfire Session
Learn how teachers use AI to write faster, more personalized student feedback. Save hours on grading & comments with Flint's AI teaching assistant.

Teddy Lane, Customer Success Engineer at Flint
In this Grading and Writing Feedback Campfire Session, we explored how teachers are using Flint to reduce the cognitive load of individualized feedback while ensuring every student receives consistent, high-quality, and actionable comments — whether for end-of-year reports, essay grading, or direct student-facing feedback.
The session combined real classroom demonstrations from two educators with live platform walkthroughs, covering strategies for synthesizing qualitative and quantitative data, iterating on Sparky's output to match a teacher's voice, and giving students the tools to engage with feedback directly.
Content covered in this session includes:
Using Flint as a teaching assistant, where teachers read student work and prompt Sparky to generate a ready-to-use, student-facing comment that can be copied, pasted, and refined.
Using Flint for direct student feedback, designing activities where students receive structured feedback without teacher intervention, such as a topic sentence review tool that guides students toward improvement without giving them the answer.
Splitting feedback work with Flint's voice tool, where teachers dictate observations on a split screen while Sparky converts those notes into a polished, robust comment in real time.
Uploading rubrics, assignments, and scaffolding materials to train Sparky to match a teacher's tone, methods, and feedback priorities before assigning to students.
Synthesizing qualitative teacher observations and quantitative grade data by uploading both into Sparky and prompting it to generate a combined, student-facing comment for each student.
Generating dual sets of comments from a single prompt, producing both student-facing and parent-facing versions with appropriately differentiated language.
Iterating on activity behavior inside Flint by adjusting Sparky's instructions mid-build to fine-tune output, with the simulate session panel available to test changes before going live.
Accessing class-wide feedback data including strengths, areas of improvement, and suggested follow-up activities, with color-coded highlights linking directly to the moments in a student's chat where those insights were identified.
Downloading all student feedback data as a CSV file for further processing, re-uploading, or record keeping.
Christa Forster's transparency approach — explaining to students exactly how Flint is used in the feedback process and why, building trust while preserving the human element of feedback.
Helene Marquette's IB assessment preparation example, where students completed homework independently and received structured, criteria-based feedback in French language and literature, saving significant class time.
Slides from the presentation can be found here.
Got more questions, comments, or feedback for this topic? Feel free to raise them within the Flint Community.
Join our next Campfire Session 🏕️
Subscribe to our events calendar to be notified when upcoming Campfire Sessions and other Flint events get scheduled.
See Events Calendar

00:00 Introduction
Teddy Lane opens the session and welcomes attendees, noting the timing aligns with the end of the school year and the start of summer.
Noah Savitz joins to monitor the chat and field questions throughout the session.
Christa Forster, English teacher at Kincaid School in Houston, Texas, is introduced as the session's guest educator.
Helene Marquette, an IB French language and literature teacher, is introduced as an asynchronous contributor who submitted a video walkthrough of her Flint use.
The session goal is framed around how Flint can help teachers generate feedback, synthesize ideas, and make written comments feel more like a conversation with students.
03:37 Icebreaking reflection
Teddy poses the icebreaker question: think back to a piece of feedback you received that was memorable — what made it stick?
Christa shares a story about receiving feedback in 12th grade English where her teacher accused her of plagiarizing a paper for using the word "primordial," a moment that stuck with her for years as an example of feedback built on false assumptions.
Attendees contribute in the chat around what makes feedback memorable: being realistic and actionable, identifying root causes rather than surface-level observations, and using language that is accessible and appropriate for the student.
Teddy connects the discussion to Flint's role in helping teachers move away from grade-focused feedback toward comments that feel more constructive and conversational.
The question of differentiating feedback for students versus parents is raised in the chat, with Teddy confirming Flint can generate dual sets of comments tailored to each audience.
10:25 Christa Forster on using Flint to give writing feedback
Christa describes the high cognitive load of giving individualized feedback on student writing, particularly when grading 50+ papers, noting that by paper number 5 or 10, the quality and specificity of comments can start to decline.
She outlines three approaches she uses with Flint for student feedback: Flint as a teaching assistant, Flint for direct student feedback, and splitting work with Flint using the voice tool.
For all three approaches, Christa emphasizes the importance of training the activity first by uploading the assignment, rubric, and any scaffolding materials, then iterating to match her tone and feedback priorities.
In the teaching assistant approach, Christa reads student work first to form her own impressions, then pastes the student's response into Sparky and asks for a ready-to-use, student-facing comment that she copies, pastes, and refines in a Google Doc.
In the direct student feedback approach, Christa demonstrates a tool she designed for students preparing for their final, focused on reviewing and improving topic sentences, where Sparky reminds students of what strong topic sentences do and gives feedback on their practice writing without ever giving them the answer.
She notes that training Sparky never to give students the answer requires consistent iteration and reinforcement within the activity settings.
In the voice tool approach, Christa reads a student essay on a split screen with Sparky open on the other side, dictating her observations aloud as she reads, and Sparky converts those spoken notes into a polished, structured comment.
Christa stresses that transparency with students about how Flint is used in the feedback process is essential, sharing that her students appreciated understanding the training process and the rationale behind it.
She shares that students gave positive feedback about Flint-assisted comments in teacher evaluations, but also noted they still want a human touch, reinforcing why transparency and teacher involvement remain critical.
23:15 Walkthrough of using Flint for grading and feedback writing
Teddy demonstrates how to synthesize qualitative and quantitative student data using Flint, showing two documents: one with personal teacher observations per student and one with quantitative grade data pulled from a gradebook.
Both documents are uploaded into a Sparky chat, where Teddy provides context on what each document contains and instructs Sparky to create an activity that presents a synthesized comment for each student by combining the two sources.
The chat is converted into an activity using the "make it an activity" prompt, and Teddy walks through how the activity appears in the activity builder, including how to verify Sparky's behavior in the Build Manually panel.
The microphone input option is highlighted as a key time-saving tool, allowing teachers to speak observations about students rather than type them, accessed via Activity Settings → Language → Student Communication → Writing or Speaking.
Teddy demonstrates using the microphone mid-session to add a sentence to a student comment, asking Sparky to include reasoning behind a specific recommendation, and showing how Sparky integrates that rationale naturally into the comment.
The simulate session panel is shown as a way to test how the activity behaves before going live with students, without needing to start a full session.
Teddy explains how to iterate within a single activity rather than building a new one, by returning to Activity Settings → Behavior and speaking or typing adjustments directly to Sparky.
Helene Marquette's video is played, showing how she built a NYP e-assessment preparation task in French language and literature, uploading a text extract, video transcript, five questions, an essay title, and a mark scheme to Sparky so students could complete practice independently at home and receive structured, criteria-based feedback.
Helene demonstrates the class-level feedback view, including strengths, areas of improvement broken down question by question, and suggested follow-up activities, with green highlights linking to students who answered particularly well and orange highlights linking to students who struggled.
Individual student feedback is shown, including personalized strengths, areas of improvement, and follow-up activities, with Helene noting she agreed with Sparky's assessments for virtually every student.
Teddy walks through how to access individual student feedback within an activity, clicking through from the class overview to a single student's session, and showing how orange and green highlights link directly to the specific moment in a chat where a strength or area for improvement was identified.
The Download as CSV feature is demonstrated, showing how all student feedback data — including strengths and areas of improvement — can be exported from an activity into a spreadsheet for further use or re-uploading into a new Sparky activity.
Teddy shows how a downloaded CSV can be attached to an existing activity in Flint and described to Sparky via microphone so it is incorporated into synthesized student comments going forward.
57:23 Q&A session
Ian asks whether it is possible to access qualitative feedback for a single student across multiple Flint sessions throughout the year; Teddy demonstrates using the search function in Flint to pull up an individual student's profile, showing their groups, activities, sessions, and chats.
Teddy notes that while CSV download currently exists at the class activity level, individual student data download across multiple sessions is not yet available but is identified as valuable feedback for future development.
Mark asks about creating an activity to process Flint feedback into notes for both students and parents; Teddy confirms this is possible and suggests it as a topic for an upcoming smaller info session.
Teddy previews the new S'more Info Sessions format: shorter 30-minute bite-sized demos with Q&A, focused on quick wins and deep dives on specific topics like building an activity from scratch or improving an existing one.
The next Campfire Session is announced for mid-June, focused on tips and strategies for updating school AI policies to be flexible and forward-looking as AI tools continue to evolve.
Have a use case you'd like us to cover in a future Campfire? Reach out and let us know!

