Skip to content
Artifact Intelligence

From a moment in a classroom to a decision that changes it.

Six movements, from a moment inside a classroom to a decision made somewhere else in the institution — and back again. None of them require anyone to change how they teach or learn.
01 — 02
Capture and connect the signal
03 — 04
Understand and model the pattern
05
Act, with the decision left to people
06
Learn, and return to the beginning

Six steps, one continuous movement.

Described separately, these read like stages in a pipeline. In practice they overlap continuously: signals are being captured while earlier signals are being modeled, and outcomes from last term are already reshaping what the system understands about this one.

The best intelligence doesn't interrupt the experience.

  1. 01

    Capture

    Simple interactions gather meaningful signals without disrupting learning.

    Prompts are placed where an answer is actually knowable — at the moment a concept lands, or fails to. Participation costs a few seconds and produces something structurally useful: a state, a concept, a timestamp, a context.

    • No surveys, no separate workflow, no added administrative burden
    • Every signal is anchored to a concept, a moment, and an environment
    • Participation is visible to the participant and never silent
    00:1200:3400:47ONE SESSION · UNINTERRUPTED
  2. 02

    Connect

    Signals connect with context, activity, historical behavior, and institutional data.

    Alone, a comprehension signal is an opinion. Joined to attendance, submissions, curriculum sequence, advising history, and prior outcomes, it becomes evidence. Artifact reads from the systems an institution already runs rather than asking it to migrate.

    • Reads from the SIS, LMS, and data warehouse; replaces neither
    • Identity and access governed by institutional policy
    • Purpose limitation enforced in the schema, not only in policy
    SIGNALSLMSSISADVISINGOUTCOMESCONNECTEDRECORD
  3. 03

    Understand

    Patterns begin revealing how learning and behavior relate to outcomes.

    With enough connected signal, relationships surface: which conditions precede recovery, which precede disengagement, where understanding reliably forms, and where a curriculum sequence consistently produces friction.

    • Relationships examined at concept, course, cohort, and program scale
    • Findings are explainable — a person can ask why and get an answer
    • Disparate impact tested continuously rather than assumed away
    BEHAVIOUR × CONTEXT■ STRONG RELATIONSHIP TO OUTCOME
  4. 04

    Model

    The platform models possible future trajectories and pathways.

    From a current state, several trajectories are plausible. Each carries an estimated likelihood, the conditions contributing to it, and the decisions that would measurably shift it. Concepts from decision modeling help identify which of those decisions carry the most leverage.

    • Pathways, not verdicts — a set of possibilities with their conditions
    • Surfaced early enough that an intervention can still change the outcome
    • Evaluated on whether flagged trajectories improved, not on accuracy alone
    NOWMODELED HORIZON
  5. 05

    Act

    People receive intelligence they can use to make better decisions.

    The same intelligence is expressed differently for each person who can act on it — because a student, a professor, and an advisor are each making a different decision. The system informs. The person decides.

    • Delivered where the decision is already being made
    • Expressed in the vocabulary of teaching and advising
    • No consequential action is ever taken by the system alone
    STUDENTA concept worth revisiting this week
    PROFESSORWhere the cohort lost the thread
    ADVISORA conversation worth having early

    The same underlying intelligence, expressed three different ways — because three different decisions are being made.

  6. 06

    Learn

    The system continuously learns from outcomes, creating an evolving intelligence loop.

    What actually happened returns to the system as evidence. Over successive cycles, the intelligence layer becomes more specific to the institution that produced it — which is the point. A model of learning at one university should not be a model of learning everywhere.

    • Outcomes feed back as evidence, not just as records
    • The intelligence layer becomes more institution-specific over time
    • Assumptions are re-examined as the environment changes
    CYCLE 1CYCLE 2CYCLE 3CYCLE 4CYCLE 5MODEL CONFIDENCE · ILLUSTRATIVE

Experience, signal, intelligence, prediction, action, outcome.

The sequence is not a line. Every outcome becomes the next cycle's evidence, which is what makes the intelligence layer specific to the institution that produced it.
CONTINUOUSINTELLIGENCELOOPEXPERIENCESIGNALINTELLIGENCEPREDICTIONACTIONOUTCOME

What this requires of an institution.

The most common question institutions ask is what this will cost their people. The honest answer is that the design constraint runs the other way: if it costs faculty and students meaningful time, the signal will be both burdensome and unreliable.
Of students
A few seconds, a few times a week, inside an environment they are already in.
Of faculty
No change to how they teach. Prompts are placed with them, not for them.
Of IT
Read access to existing systems, governed by institutional policy and reviewed on the institution's terms.
Of leadership
A decision about what the institution actually wants to understand — which is the part that takes the most work.
A seminar room mid-discussion: adults seated around a large table, one speaking, daylight from a window wall with a campus building beyond.
A workshop wall at the end of a design session: a large hand-drawn diagram of connected boxes and a numbered sequence, flanked by columns of sticky notes.

What would this look like in your environment?

Every institution's version of this sequence looks slightly different. The first conversation is usually about which questions are worth answering.

Talk With ArtifactExplore Solutions