Skip to content

Tools: the three pillars

AdaptiveLearner doesn't try to be your only learning tool. It tries to be the orchestrator that points you at the right external tool for what you're doing right now. Five tools ship in the catalogue, mapped to three pillars.

The three pillars

1. Spaced repetition

The cognitive science is settled: spaced review beats massed review for long-term retention. The interval between reviews matters; tools like Anki turn this into a discipline.

Recommended tool: Anki. Free on desktop, paid on iOS, the schedule algorithm is well-tuned, and it's the de-facto standard. Most other apps in this space copy Anki's intervals.

Use it for: anything that has to be remembered long- term. Vocabulary, formulas, named entities, error-correction recipes. AdaptiveLearner sessions are great for understanding; Anki is great for not forgetting.

The Adaptive Learner profile weights "deductive" and "error_based" most strongly toward Anki, since both methods produce material that's worth carding.

2. Active recall from your own sources

The second pillar is building knowledge from documents you provide. Modern tools let you upload PDFs, notes, transcripts and then quiz yourself or ask questions grounded in that specific corpus.

Recommended tool: NotebookLM. Google's tool that turns your sources into an interactive knowledge graph. Better than ChatGPT for this purpose because the AI's answers are anchored in your provided documents.

Also useful: Excalidraw for sketching out the structure of your knowledge, Obsidian for linked-notes knowledge graphs that grow over months.

Use it for: domain-specific learning where you have existing material (research papers, internal docs, course slides) and want to extract knowledge structure from it.

3. Adaptive AI prompts

The third pillar is direct AI access for one-off questions that don't fit the other two pillars. Sometimes you just need an explanation. Sometimes you want to brainstorm.

Recommended tool: Claude, ChatGPT, or Gemini. AdaptiveLearner uses the same APIs internally; you can also talk to them directly in their web UIs for less-structured exploration.

Use it for: open-ended questions, brainstorming, "explain this paragraph", "give me three different framings of this problem". The unstructured chat shines for divergent thinking; AdaptiveLearner sessions shine for focused convergent practice.

How AdaptiveLearner ranks tools

The Dashboard's Tool Recommendations card runs a simple scoring:

score(tool) = sum(profile_weight[k] for k in tool.weight_keys)

Each tool declares which 1-2 method axes it best serves:

Tool Weight keys Why
Anki deductive, error_based Cards encode rules + corrections
NotebookLM inductive, contextual Examples + situated material
Adaptive AI Prompt ai_adaptive, dialogic Adaptive conversation
Excalidraw contextual, inductive Visual structure from examples
Obsidian deductive, inductive Theory + examples in one graph

Profile weights from your assessment are summed across each tool's weight_keys, and the ranked list is what the Dashboard surfaces. The ranking updates whenever your profile changes (re-evaluating the assessment).

Spaced recommendations

A second tracking surface on the Dashboard: the Spaced card. This is NOT tool recommendations; it's action recommendations. The system tracks how long it's been since you last had a session in each method, then suggests:

Time since last commit Card kind Interval
Never first 1 day
> 14 days refresh 1 day
7-14 days review 3 days
3-7 days practice 7 days
< 3 days maintain 14 days

So a method you haven't touched in two weeks gets a "Refresh in 1 day" card. A method you used yesterday gets "Maintain in 14 days" (or doesn't surface at all because the list is capped to 5).

The cards are sorted by urgency (lower interval × stronger weight = higher priority). You don't have to follow them — they're nudges, not commands.

What's NOT in the catalogue

Deliberately excluded:

  • Duolingo / Babbel / similar gamified apps — they conflict with the philosophy. AdaptiveLearner is un-gamified by design.
  • Khan Academy / Coursera — they're course-completion oriented, not skill-acquisition oriented. Different problem space.
  • Memrise — too close to Anki; the catalogue keeps one tool per niche.
  • Notion — overkill for the "linked notes" niche; Obsidian fits cleanly without a cloud lock-in.

The catalogue is small on purpose. Five tools cover the space. Adding more would dilute the signal.