Ongoing Research · Methodology

Research Methodology

How we approach two complementary frontier areas — AI Personalization and InPlaceAI — with a shared commitment to individual context, seamless integration, and responsible augmentation of human capability.

Explore the Pillars Partner with Us
Two Research Pillars

Personalization & InPlaceAI

Two complementary research directions that share a common methodology — but apply it to different frontiers of human–AI interaction.

AI Personalization

Deeply Individualistic Intelligence

AI that is deeply individualistic — tied to the individual’s style, preferences, and context. Moving beyond one-size-fits-all to truly personalized intelligence. Our research explores how models can adapt to the individual without losing general capability.

Active Research

InPlaceAI

Right Where We Work

Creating AI that is right where we work every day — seamlessly integrated into your workflow, anticipating needs, and augmenting human capability without disruption. Our research explores ambient, context-aware AI that disappears into the tools people already use.

Research in Progress
Shared Methodology

Common Principles

Both research pillars are guided by the same methodological commitments — even as they explore different frontiers.

Individual Context First

Every design decision starts with the individual’s style, preferences, and working context.

Augmentation, Not Replacement

AI should augment human capability without disruption — never substitute for human judgment.

Privacy-Preserving Adaptation

Personalization and context-awareness must respect privacy — local-first where possible.

Iterative & Evaluative

Continuous cycles of prototyping, user testing, and refinement across real workflows.

Ethics & Transparency

Explainable behavior, user control, and ethical review embedded from the start.

Measurable Impact

Success measured by real productivity, satisfaction, and reduced friction — not vanity metrics.

Pillar-Specific Methodology

How We Research Each

While the principles are shared, each pillar applies them to distinct research questions and technical challenges.

AI Personalization

Methodology Focus

We treat personalization as a research problem in adaptive model behavior — how to bind intelligence to an individual’s style, preferences, and context without sacrificing generality or privacy.

  • User modeling research: Representing individual style and preference signals in ways the model can act on.
  • Adaptation mechanisms: Exploring lightweight personalization layers (LoRA-like adapters, prompt tuning, retrieval-augmented context).
  • Privacy-preserving learning: Local-first adaptation, federated approaches, and user-controlled data boundaries.
  • Evaluation: Measuring personalization lift against a generic baseline — without regressions in core capability.
  • Controllability: Users can inspect, adjust, and reset their personalized profile at any time.
Active Research

InPlaceAI

Methodology Focus

We treat InPlaceAI as a research problem in ambient, context-aware integration — how AI can live inside existing workflows, anticipate needs, and augment human capability without adding friction.

  • Workflow observation: Ethnographic and instrumentation studies of real day-to-day workflows to locate high-value intervention points.
  • Context sensing: Exploring signals (documents, messages, calendar, app context) that allow the model to anticipate needs without intrusive capture.
  • Seamless integration patterns: Embedding AI in existing tools and interfaces rather than requiring new user habits.
  • Non-disruptive augmentation: Designing suggestions and automations that can be ignored, overridden, or refined without friction.
  • Evaluation: Measuring reduced friction, time saved, and user trust — not just task accuracy.
Research in Progress
Where the Pillars Meet

Combined Research Focus

Personalization and InPlaceAI are not silos — the most interesting questions sit where they intersect.

Individual Context + Workflow Context → Context-Aware Personal AI → Non-Disruptive Augmentation

Personalized Workflow Intelligence

AI that understands both who you are and what you’re doing — adapting to your style while remaining ambient in your workflow.

Anticipation Without Intrusion

Using individual and contextual signals to anticipate needs — while keeping the user in control of when and how AI intervenes.

Friction-Minimizing Personalization

Personalized adaptation that requires no extra effort from the user — integrated into the tools they already use, in the way they already work.

Privacy-Preserving Context

Handling both personal and workflow context with strong privacy boundaries — local-first where possible, transparent always.

Collaborate on Our Research

Interested in partnering on AI Personalization or InPlaceAI research, or exploring how these directions could apply to your workflows? Let’s talk.

Talk to Our Team Back to AI Research