Tailor this document
Fact Sheet · 2026
AlwaysAI

AI-Integrated Assessment & Learning Platform — where AI collaboration is mandatory, visible, and graded using a dual-scoring architecture (O×A framework).

Live at
alwaysai.skoonline.org
Part of
skoonline.org ecosystem
AlwaysAI flips the AI-in-education problem: instead of banning, ignoring, or detecting AI use, it assesses how well students collaborate with AI. Every assessment produces a dual score — outcome quality and AI collaboration quality — because students learn what they are assessed on. The result is a generation of graduates who use AI critically, not blindly.
Our philosophy

Most educational institutions treat AI as either a threat to police or a tool to ignore. AlwaysAI is built on a different premise: if you want students to use AI correctly, assess their AI usage directly. Every design decision flows from six beliefs about how AI literacy actually forms.

Constructive Alignment
Students learn what they are assessed on — so assess AI collaboration directly
Traditional assessments tell students “don’t use AI” or silently allow it with no feedback loop. AlwaysAI makes AI collaboration a graded dimension of every task. When students know their AI usage quality is measured, they develop genuine collaboration skills rather than superficial copy-paste habits.
Dual Scoring · O×A
One score is never enough — separate outcome quality from AI process quality
A brilliant final product produced by mindless AI copy-paste is fundamentally different from the same product produced through skilled AI collaboration. AlwaysAI separates what you produced from how you worked with AI to produce it, and weights them according to the assessment philosophy. Both dimensions matter.
Transparency Over Surveillance
Teachers and students see the same view — grading is a conversation, not a verdict
Every rubric, every AI dimension, every weight is visible to the student before, during, and after the assessment. There is no hidden scoring. Students can revisit their work, see exactly where they lost points, and improve. Assessment becomes a learning tool, not an opaque judgment.
Multi-Session Journeys
Learning is not a single event — it unfolds across sessions, days, and phases
Real learning happens over time: exploration, struggle, revision, insight. AlwaysAI supports multi-session learning activities that track process quality across days or weeks — not just the final artifact. Phase progression, session spacing, and knowledge construction are all measured.
Workflow-Driven Assessment
Assessment is not “answer a question” — it is a structured professional workflow
Every assessment includes explicit workflow steps, key questions, and common mistakes. Students work through a guided process — drafting, refining with AI, iterating, synthesizing — that mirrors how professionals actually use AI in their work. The workflow itself is part of what’s evaluated.
Provider-Agnostic AI
The skill is AI collaboration — not proficiency with one vendor’s product
AlwaysAI supports 11 LLM providers. Students develop transferable AI collaboration skills — prompt precision, critical evaluation, integration — that work regardless of which model sits behind the interface. No vendor lock-in. The competency is universal.
Theory behind AlwaysAI
🎯
Constructive Alignment
Biggs, 1996
Learning outcomes, activities, and assessment must be aligned. If AI collaboration is a desired outcome, it must be assessed.
🧱
Constructivism
Piaget · Vygotsky
Knowledge is actively constructed. AI becomes a scaffold for building understanding, not a substitute for thinking.
🎓
Bloom’s Taxonomy
Bloom et al., 1956
AI collaboration dimensions map to higher-order thinking: Analyze (critical evaluation), Evaluate (metacognition), Create (synthesis).
📊
Learning Analytics
xAPI · Evidence-Centered Design
Every interaction becomes evidence. Stealth assessment via xAPI captures process, not just product.
Full overview →
11
LLM Providers
OpenAI · Anthropic · Google · +8
6
AI Dimensions
measured per interaction
4
Assessment Philosophies
configurable scoring weights
20
Learning-Science Structures
Inquiry · 5E · Productive Failure · +17
9
Languages
incl. RTL (Arabic, Hebrew)
Who it’s for
🎓
Students & Learners
Collaborate · Iterate · Synthesize · Grow
📋
Teachers
Design · Configure · Monitor
  • Create assessments with structured workflows, key questions, and rubrics
  • Choose scoring philosophy per course: professional, formative, certification, or AI proficiency
  • Build multi-week module curricula with learning activities and assessments per week
  • Monitor class-wide AI collaboration patterns and individual student progress
  • See the same view students see — full transparency in grading
  • Generate class reports with diagnostic patterns and recommendations
🏛
Institutions
Integrate · Scale · Comply
  • xAPI / LRS integration — every interaction logged as a learning record
  • OAuth sign-in via Google (extensible to Microsoft, GitHub)
  • Multi-institution support with school-level admin roles
  • Google Cloud Storage for persistent data and backups
  • Role-based access: student, teacher, school admin, super admin
  • Deploy to App Engine, Cloud Run, or Docker — your infrastructure
🔬
Researchers
Measure · Analyze · Publish
  • Rich xAPI data on student-AI interaction patterns at scale
  • 6-dimension AI collaboration scoring with configurable weights
  • Longitudinal data from multi-session learning journeys
  • 5 AI role classifications per interaction (executor, retriever, interlocutor, critic, generator)
  • LLM analytics: provider, model, tokens, latency, cost per interaction
  • Open scoring framework adaptable to new research questions
Assessment & learning modes
📝
Structured Assessment
Single-session evaluation with workflow steps, AI collaboration, and dual O×A scoring
📖
Learning Activity
Multi-session PBL, project-based, or case-based journeys with phase tracking and process scoring
🎬
AI Lectures
Slide-by-slide narrated lectures with TTS, embedded quizzes, and completion tracking via xAPI
🧪
Trial Assessment
Low-stakes practice mode for students to experience the workflow before graded assessment
🗣
Voice Mode
Real-time voice AI interaction via OpenAI Realtime API — speak naturally, get scored identically
📎
Multimodal Upload
Upload images, PDFs, CSV, or text files as part of AI interactions — supports visual reasoning tasks
💬
Discussion New
AI-generated multi-voice round-table discussions on any concept, with 3D cast visualization and narration
📊
Results & Diagnostics
Detailed score breakdowns, diagnostic patterns, performance tiers, and actionable improvement recommendations
How it works
01
Teacher configures
Create an assessment with workflow steps, rubrics, AI role expectations, and choose a scoring philosophy.
02
Student works with AI
Follow the structured workflow: draft, consult AI (chat/voice/upload), iterate, synthesize. AI collaboration is mandatory and visible.
03
System evaluates both
AI scores outcome quality AND collaboration quality across 6 dimensions. Every interaction is logged to xAPI.
04
Everyone sees the same view
Student and teacher see identical results: scores, rubric breakdowns, diagnostic pattern, and specific improvement recommendations.
Key differentiators
O×A
Dual Scoring
Traditional assessment

Grade the final product only. AI use is either banned, ignored, or penalized. No signal about collaboration quality.

AlwaysAI O×A framework

Score outcome AND AI collaboration independently. Weight them per course philosophy. A great product with poor AI process scores differently than the same product with skilled collaboration.

6D
Measurable AI Skills
Typical AI literacy programs

“Use AI responsibly” — vague advice with no measurement, no feedback, no progression tracking.

AlwaysAI 6 dimensions

Prompt Precision, Iteration Depth, Critical Evaluation, Integration, Metacognition, Efficiency — each scored 0–100, weighted, tracked over time. Students see exactly where to improve.

Process
Multi-Session Journeys
One-shot assessment

A single exam measures a snapshot. No visibility into how understanding develops, how AI collaboration matures over time.

AlwaysAI learning activities

Multi-session journeys across days or weeks. Phase progression, session management, engagement depth, reflective practice, and knowledge construction — all scored. Process matters as much as product.

Transparent
Same View for All
Traditional grading

Students submit and wait. Rubrics are opaque. Feedback arrives days later with no actionable path forward.

AlwaysAI transparency

Teacher and student see identical views. Rubrics visible before, during, and after. Diagnostic patterns explain results. Students can revisit and understand exactly where to improve.

Technology
React 18 + Vite
Node.js + Express
11 LLM Providers
OpenAI Realtime (Voice)
Google Cloud Storage
xAPI / LRS
OAuth (Google)
i18next · 9 languages
Docker / Cloud Run / App Engine
Tailwind CSS
WebSocket (voice relay)
Multimodal (images, PDF, CSV)