Askhaven AI reads every learner message an online-learning team handles — chat, email, LMS help, course Q&A — grounds the answer in your own course materials, and routes anything that still needs a human to the right desk with a confidence score attached.
Reads the LMS docs, syllabi, lesson materials and FAQs you already keep — grounds every answer in a real source, cites it back to the learner, and escalates only what the model isn’t confident about.
Askhaven AI watches every message an online-learning team receives — chat, email, LMS help, course Q&A — grounds an answer in the syllabus the learner is actually enrolled in, and shows the confidence, the routing decision and the escalation reason side by side. Support agents work a ranked queue, not an inbox.
Routing, Q&A, ticketing, scheduling, content generation, learner analytics and document processing all read from the same learner record — so a support agent sees the same evidence the routing engine used, and no answer has to be re-keyed between teams.
The queue reads each incoming learner question, tags its intent — Course Help, Account Help, Policy, Technical, Other — assigns a priority, and routes it to the right support queue or automated assistant, with the routing confidence shown beside the destination.
The reviewer opens the learner's question beside the AI answer, the source it was drawn from — Lesson, PDF, Video, FAQ — the topic and course, and the confidence score, then approves, edits or escalates before it ships back.
Tickets are classified — Course, Account, Billing, Technical, Access — prioritised against the applicable SLA, and offered to the right agent or team, with the urgency rationale shown alongside the ticket history.
The calendar shows session type, learner, course, requested time and availability state — Available, Limited, Full, Conflict — and proposes an alternative when a slot doesn't work, so agents confirm or reschedule without email ping-pong.
FAQ, explanation, quiz, summary or announcement drafts are generated from approved course knowledge, targeted at a specific audience or proficiency level, and shepherded through Draft → Review → Approved → Published in one form view.
Recommendations, per-learner analytics, content discovery, document processing and the coverage map that tells you where your knowledge has gaps.
Each recommendation carries a learner, the recommended course or resource, the reason — Skill Gap, Progress, Interest, Prerequisite, Goal — a 0-1 relevance score and an expiry, so advisors approve, edit or dismiss the suggestion instead of guessing at it.
Progress, engagement, recent activity, support-need level — None, Guidance, Intervention, Escalation — and intervention history sit on one record view, so an advisor sees why a learner has stalled without stitching four reports together.
A semantic index across every lesson, video, PDF, quiz, FAQ and assignment returns the best-match resource for a learner or support ticket — with topic, type, access level and a 0-1 relevance score — so agents recommend the existing answer before anyone writes a new one.
PDFs, DOCXs, PPTXs and images are parsed side by side with the original — extracted fields carry a confidence score and a Good / Review / Poor quality state, and the workflow moves each document through Uploaded → Processing → Review → Approved.
A coverage heatmap over FAQs, LMS documentation, syllabi, lesson materials and assignment guides — with answer rate, citation rate and indexed passage counts on every row — so you know which source is thin before the escalation queue tells you.
Low-confidence conversations are routed to the right desk automatically — with the escalation reason and handoff confidence carried onto the ticket.
A conversational front door for LMS help, deadlines, billing and course questions — grounded in your own materials and cited back to the learner.
Practice questions, quizzes and formative checks are drafted from a lesson's own knowledge and targeted at a specific proficiency band.
Written feedback on submitted work is drafted, tagged Positive / Constructive / Negative, and returned to the instructor for review.
Enrollment forms, IDs and credentials are read field by field beside the source, with confidence scores and a Valid / Needs Review / Missing Data state.
Learners at risk of falling behind are surfaced with the evidence — activity drop, missed assessments, engagement level — and a prioritised list of interventions.
Connect the LMS, the syllabi, the lesson PDFs, the FAQs and the ticketing tool you already run — nothing is re-keyed, no new data entry lands on your team.
Each learner question is answered from a real source in your course library — with the citation shown to the learner and the confidence shown to the reviewer.
Only the low-confidence and off-course conversations reach a human — ranked, classified, and prioritised — so the queue is short and the tickets already carry context.
78%
Of conversations resolved without a human
18s
Median first response, across every intent
92%
Answer confidence on every routed message
94%
Routing accuracy on LMS-access questions
Askhaven AI is built around the regulations that govern learner conversations, accessibility, minor consent and cross-border data — here is what it supports out of the box.
Student records are handled as protected education records with role-scoped access and audit trails on every enrolment, credential and progress transaction.
For learners under 13, the support chatbot honours verifiable school or parental consent, plus the 2026 rule updates on third-party sharing and biometric data.
The learner chatbot, Q&A widgets and content surfaces meet all 25 Level AA success criteria — including the nine new 2.2 criteria for cognitive and low-vision users.
An independent audit of the controls protecting learner conversations, transcripts and enrolment documents in transit and at rest across the platform.
Lawful-basis, data-subject-rights and cross-border-transfer obligations for EU learners are enforced across analytics, chatbot transcripts and document processing — with DPA and SCCs available.
$200/mo
Flat for the whole platform, billed monthly. Cancel or change any time — every conversation transcript exports with you.No seat counting as your learner base grows and no feature held back for a higher tier. Talk to us and we’ll shape it around your course catalogue and support volume.
A one-time setup that gets your platform live before the first recurring cycle — we ingest your course materials, connect your LMS and ticketing tool, and tune the routing, grounding and escalation rules to the way your support team already runs. Each fee below is a starting point: the final onboarding is scoped once we understand your environment, security needs, integrations, data volume, compliance requirements and deployment model.
Everything in Essential, plus:
Everything in Plus, plus:
Same recurring subscription either way — only the onboarding may differ by your requirements.
Runs in Nestack's managed cloud — nothing to provision, updates and scaling handled for you. The quickest path from onboarding to your first live conversation.
The same platform deployed inside your own infrastructure — your cloud, VPC or on-prem — for teams handling sensitive learner data. Chatbot transcripts, enrolment documents and analytics run where your FERPA / COPPA / GDPR-governed data already lives, and the recurring subscription stays the same.
It handles the learner-side support your platform receives — LMS access, deadlines, billing, course Q&A, scheduling — grounded in your own syllabi, lesson PDFs and FAQs. Every message is classified, given an intent and priority, and either answered by the bot with a citation or routed to the right desk with the confidence score attached.
The course library and knowledge base you already keep — syllabi, lesson PDFs, FAQ articles, LMS documentation and video transcripts. The Q&A bot only answers from a real source and cites the passage back to the learner; anything it isn't confident about is escalated rather than made up.
Containment is the share of learner conversations resolved without a human ever touching them. The dashboard reports it live against your target (default 75%) and breaks it down by intent, so you see whether LMS-access questions or deadline queries drive the escalations.
Low-confidence conversations don't get shipped to the learner. They land in the reviewer's queue with the draft answer, the source it was grounded in, and the escalation reason — the reviewer approves, edits or hands them off to the right agent.
Yes. Nothing is auto-resolved above the confidence threshold you set. The router ranks and prioritises; agents accept, reroute or resolve. The dashboard tracks that split so you can see how much of the queue your team is actually acting on.
Online learning platforms, MOOC providers, ed-tech companies and any team running courses at scale — anywhere there is a learner base, a course catalogue and a support inbox worth deflecting.
Book a demo and pick the modules you want to see. We’ll walk Askhaven AI through a slice of your real support volume — routing, grounded answers, the escalation queue and the coverage map, end to end.
Tell us who you are, pick what you want to see, then grab a time — we’ll run Askhaven AI on a slice of your real learner conversations.
Takes a few seconds — so we come ready for your platform.