What's next

The research ends.
The question doesn't.

The Applied Research Project is a starting point. These are the open questions, the future validation work, and how to stay connected if this matters to you.

Honest scope

What this research does not resolve.

The ABSD framework is a conceptual contribution. It organises existing research and proposes a four-layer architecture for understanding user motivation in AI systems. It does not, in this study, prove that the framework produces better outcomes when implemented. That validation requires a different kind of research.

The practitioner survey and expert interviews validate that the gap ABSD describes is recognised - that practitioners see themselves in the framework. They do not validate the framework's inferential accuracy, its production implementation requirements, or its effectiveness across domains beyond CX and EdTech.

The framework is a map. The map has not been tested against the territory at scale. That is the work remaining.

Open questions

Seven questions this research leaves unanswered.

01
Cross-domain stability. The ABSD layer architecture was developed with CX and EdTech as primary domains. Do the same four layers and seven signals hold in healthcare technology, financial services, and enterprise software? Or does the framework need domain-specific modification?
02
Inferential accuracy in production. Affective computing achieves 70–79% accuracy in controlled settings. What does that number become in a real deployed system at scale, with noisy data, varied user populations, and no ground truth labels?
03
Drive inference feasibility. SDT provides the theoretical constructs for Layer 4. The longitudinal observation requirement is well-established. What is the minimum data history required to produce a reliable drive inference? Weeks? Months? Does it depend on interaction frequency?
04
Intervention outcome evidence. JITAI research shows state-aware interventions outperform static ones. Does ABSD-informed intervention - drawing on all four layers - outperform single-layer JITAI? What is the marginal value of each layer?
05
The privacy-utility tradeoff. Deeper inference requires more data and more longitudinal observation. At what point does the privacy cost of Layer 3 and 4 inference exceed the user benefit? Who decides?
06
Drive stability over time. SDT describes drives as relatively stable psychological needs. But people change. Career transitions, life events, and repeated frustration all affect motivational profiles. How does a drive model handle drift?
07
The agentic AI question. The ABSD framework was designed for systems that respond to users. As AI systems become more agentic - acting on behalf of users rather than responding to them - does the framework still apply? Or does it need a fifth layer?
Future research directions

What comes after this study.

Short
Cross-domain validation study
Replicate the practitioner survey with a healthcare technology cohort and an enterprise SaaS cohort. Test whether the seven signals and four layers generalise beyond CX and EdTech.
Short
Signal detection prototype
Build a lightweight Layer 2 signal detection library - open source - that product teams can apply to their existing event logs without new infrastructure. Validate the taxonomy against real data.
Medium
Layer 3 intervention study
Partner with an EdTech or CX platform to run a controlled comparison: ABSD-informed state-aware interventions versus standard rule-based triggers. Measure outcomes over a full user cohort for one quarter.
Medium
Drive inference methodology paper
A focused methodological contribution: how to operationalise SDT's three basic psychological needs as inferences from interaction data. What signals, what time windows, what accuracy thresholds are required.
Long
ABSD as an industry standard
If the framework validates across domains, the long-term goal is adoption as a shared vocabulary for behavioural intelligence in AI systems - the way the OSI model gave networking a shared architecture. That requires community, not just research.
Stay connected

Get the final version before it goes public.

Newsletter subscribers get the final research version first - before it is published publicly. No more than one or two emails total. No marketing. Just the research when it is ready.

Research updates only.
Final paper release · major framework revisions · validation study results. Nothing else.
✓ You are on the list. Final paper comes to you first.