Old oaks and moss at the edge of a field near Nashville, Tennessee

An invitation

The guide asks. The student thinks.

Socratic Guides is a small academy now in formation in Franklin, Tennessee. A trained human guide and an AI companion will work alongside each student — not to hand over answers, but to keep asking until the student can defend one. Music and time outdoors will be part of the schedule, not extras.

Read the method Schedule a conversation

The plan, in the founder's voice

Before the method or the budget, there is the essay: the founder's plain statement of what the Academy is for, how it treats a child, and why it asks corporations to back it.

Read the plan

Who this is written for

This site is written for parents evaluating the program and for sponsors and investors evaluating the case. Students are addressed in the sessions themselves, at full strength: no simplified version of an idea, and a guide whose job is to ask what the student actually thinks about it.

The work is founded by an American citizen-scientist: someone who builds the instrument, runs the experiment, publishes the failure, and revises. That posture is the curriculum before it is the pedagogy. We are not an institution asking for deference. We are a small group asking testable questions in public.

The deeper written work — books, essays, and the research archive — is being collated and indexed, and is not yet published alongside this site. This site covers the Academy only.

Get pilot updates

One short email when the pilot cohort opens, when a finding is published, or when a visit date is scheduled. No other use of your address.

We store your email address and nothing else you did not type here.

Six commitments

Questions before answers

Every session begins with a question the guide does not answer. The AI companion is configured to withhold conclusions and return structure instead.

Adult language, honest difficulty

No mascots, no simplification theater. We adjust pace and scaffolding for a young mind; we do not lower the ceiling of the idea.

Music and outdoor time as structure

Instrument practice and time outdoors are scheduled blocks with defined cognitive purpose, not enrichment padding.

Claims carry status labels

Everything we assert about learning outcomes is marked as hypothesis, pilot finding, or replicated result — with the observation that would falsify it.

Small by design

The first cohort is capped so that each student is known by every guide. Growth follows evidence, not demand.

In formation, stated plainly

We do not yet have a licensed school, a campus, or a secured site. No private residence is a school facility, and we will not describe one as though it is.

What we are building toward

Architectural rendering of a timber-framed learning room: skylit vaulted ceiling, curved tables with woven floor cushions, reference screens along a low bench, glass wall onto a planted courtyard, and a raised alcove holding hand drums and small harps — A rendering of the instruction room we are working toward: one shared table instead of rows, instruments in the room rather than down the hall, and daylight and planted ground visible from every seat. It is a drawing of an idea — not a building we have, not the existing house, and not a plan we have funding or approval to construct.

See where the Academy stands on location and facilities, and what the drawing is arguing about arrangement.

What we are currently claiming

Hypothesis — not yet tested

Pairing melodic encoding with Socratic recall may improve durable memory retrieval relative to conventional review. A figure of roughly 41% improvement has circulated from small external studies; we treat it as an untested hypothesis for our setting, not a finding of ours. It is falsified if matched-cohort delayed recall shows no reliable difference.

Pilot finding — provisional, n is small

Students given a question rather than an explanation produce longer and more self-corrected reasoning in written work. Provisional, unblinded, and small-sample. It is falsified if blind scoring of matched writing samples shows no difference.

We publish the falsifier alongside the claim, every time.
We publish results that go against us at the same prominence as results that favor us.
We do not describe an outcome as caused by our method until it has replicated outside our own cohort.