loading...
Packback uses AI to spark genuine curiosity – peer-driven discussion boards and writing feedback that develop student voice instead of replacing it.
ENTRY ANGLES
AI tools that scale instructor feedback quality and volume to students · AI systems that reduce repetitive administrative tasks for teachers and students · AI tutoring that develops independent thinking rather than completing assignments
VERTICALS
CAPABILITIES
Educational content understanding and pedagogical design, Natural language processing for feedback generation, Student engagement and learning outcome measurement
Packback's mission is to make every student "fearlessly curious and able to find their own voice."
When a student's job is to memorize what the teacher says and play it back on demand, there's not much room for curiosity or independent thought to develop.
Packback addresses this with two AI-powered tools built on the same underlying technology.
Packback Questions sets up a class discussion board where students pose open questions to their peers – on topics they want to understand more deeply or debate more seriously. Left unmoderated, any discussion channel degrades into noise; Packback's AI moderates for educational value rather than tone. A student who tries to post “What is osmosis?” gets pushed to rephrase as an open question that invites varied, reasoned responses – even rejected posts become part of the learning process.
The AI also scores every message, weighing citation formatting alongside subtler signals like intellectual curiosity embedded in the question. Beyond moderation, it acts as a teaching assistant: commenting publicly on strong and weak contributions, tracking participation against instructor-set minimums, and generating weekly summaries of class activity with engagement scores – all without the instructor lifting a finger.
Packback Deep Dives handles longer-form work: essays and research papers. Students get automated feedback as they write or after upload; a sourcing advisor rates cited sources for credibility and auto-formats citations in academic style. On the instructor side, the same AI that coached the student during writing now suggests evaluation criteria and scores at grading time – using the same rubric applied during the writing phase, so the feedback loop is consistent end-to-end.
The result: using Packback, an instructor can structure meaningful class activity and draw every student into it – while spending their limited time on the substance of what students are thinking rather than on correcting the mechanics of how they're writing.
The data supports this. Student grades measurably improve when they use Packback – which is presumably why 600 schools and universities across the US and Canada have adopted the platform.
Packback can also be used independently by individual students, even if their institution hasn't officially deployed it.
Packback has raised another $1.5M, bringing total funding to $12.2M.
The simplest use of AI in education is letting students use ChatGPT to complete assignments for them: solve the problem, write the essay. Simple – but of zero educational value. Arguably negative value.
Packback argues that AI's role in education should be instructional, not substitutional. AI shouldn't do the work for students. It should teach them how to do it – directing their thinking toward the right process.
This immediately calls to mind Socratic method – the teaching approach Socrates described this way: he couldn't teach anyone anything; he could only make them think. His method was a structured sequence of questions that led students to arrive at conclusions themselves. Whether those conclusions were what Socrates intended or what the student genuinely believed is almost beside the point – what mattered was that the student had fully internalized and owned the thinking.
Socratic method came up in a [review from summer 2023](/review/ii-vzorvjot-obrazovanie-sovsem-s-drugoj-storony) covering Sizzle, which raised $7.5M in its first round. Sizzle built a homework help app that refuses to solve problems for students – instead offering step-by-step hints that lead them to the answer themselves. That's both Socratic and structurally identical to Packback's underlying concept.
A second problem in traditional education is feedback volume. Feedback – not knowledge transmission – is the most important element of learning. But delivering regular, substantive feedback to every student at the scale they need is beyond what most teachers can do. There isn't time. There often isn't money.
AI opens up a path to solving this at scale – but not by handing the feedback process entirely to the machine. Fully automated feedback risks becoming a hollow set of generic comments that carry none of the instructor's own educational perspective.
Graide, [covered here](/review/pomogi-im-masshtabirovatsja) and carrying £2.5M in funding, addresses this elegantly: the AI generates feedback based on the teacher's own past comments on similar work. In any subject, students tend to make the same mistakes repeatedly. And a teacher who has been teaching the same material for several years has effectively built a library of feedback that can be automated against almost every variant of student response.
Packback operates on the same principle: the AI handles formal and structural feedback in the discussion thread and on submitted work, leaving the substantive response to the instructor. Even that division saves real instructor time and guarantees immediate reactions on the routine elements.
The third traditional problem is boredom. When students are bored, material doesn't land.
AI can help instructors make learning more engaging – generating compelling lesson presentations or discussion questions that spark curiosity. That's the territory of Curipod, [covered previously](/review/ne-gonjajsja-za-odinochkami-srazu-zahodi-na-rynok), a Norwegian startup that helps school teachers make their lessons more interesting and inspiring. It raised $4.8M in its first round to fund US expansion.
Packback addresses engagement differently: not through content delivery but through structured peer discussion on the topics students are already studying.
The broad direction: build platforms and tools that embed AI meaningfully in educational processes.
The pain points to target:
- Increasing student engagement and curiosity in the learning process,
- Scaling the volume and quality of feedback students receive from instructors and from peers,
- Reducing the time students and teachers spend on repetitive, formal tasks,
- Developing students' capacity for independent thinking and reasoned conclusions.
The critical constraint: AI shouldn't be used to replace both the student and the instructor – one AI completing assignments while another AI grades them. That path ends with one AI educating another.
AI in education should work the way a calculator works in mathematics – eliminating the manual arithmetic so students and teachers can focus on the interesting questions: what to compute, why it matters, and whether a different kind of math might solve the problem better.
The AI-in-education space is still young, which means there's real room for new approaches. The startups mentioned here have sketched several of the available directions – and any of them could be developed further. If no new directions come to mind, developing what's already been mapped out is still a very large opportunity.
What other uses of AI in education would genuinely raise its value and effectiveness?