Designing the expert desk: the telehealth model
A design proposal for an artifact-first consult service. The workflow is in product preview; live booking activates only after the founding roster has coverage.
Stuck is a scheduling problem
Clinical triage offers a useful analogy, not a claim that learning and medicine are equivalent: route a bounded problem, with prepared context, to the right specialist. Many online programs instead leave a stuck learner with asynchronous community support and no defined response window. The expert desk is a proposal for a different operating model, not a live service or measured learner outcome.
Consider a composite training scenario: two approaches both seem defensible, an error resists search, or a design decision has no local feedback. That may be a judgment gap rather than a missing fact, making it a candidate for a short expert consult. Whether the planned desk resolves such cases quickly is a hypothesis the service must measure once roster coverage and bookings are real.
Triage is the unsolved half. 'An expert' isn't enough — you need the one who has hit this exact wall. The planned desk therefore routes through the graph: the node determines the discipline that should receive a future request, the same way symptoms determine the specialist.
Why test a fifteen-minute unit
Search assumes you know what to ask. Stuck people don't — that's what being stuck means. An expert can invert the problem: you show the symptom, they name the disease. The service thesis is that pattern libraries take years to build, so the right practitioner may be able to identify a familiar failure from a small amount of evidence.
This is the telehealth-inspired hypothesis applied to expertise: the scarce resource may be diagnosis rather than information. Fifteen minutes is the proposed learner-facing unit, not a proven optimum. Some problems will need longer work or a different service entirely. An activated pilot would need to measure routing accuracy, resolution, follow-up, and the share of cases that do not fit the unit before making an efficiency claim.
The handoff is the product
A cold consult can spend much of its clock on 'walk me through what's going on'. The planned workflow packages the context before any future call: the node, submitted artifact, exact error, and prior attempts. The expert would review that bundle before joining so the conversation can start at depth.
The handoff also asks the learner to state what they tried and where it broke. That preparation may resolve some stalls before a booking and should make the remaining request easier to route. Both are design hypotheses to test, not results from a cohort that has already run. Rubber ducks everywhere approve.
The planned consult starts with prepared context. The pilot has to prove that the handoff improves the conversation.
The economics of micro-consults
The service thesis is that the alternative to paid specialist access may be no dependable access at all. The planned model is a paid, time-boxed slot at the expert's own rate so both sides arrive prepared and the work is compensated rather than treated as an inbox obligation.
For the expert, the proposed unit is twenty minutes of calendar — fifteen with the learner and five reviewing the handoff. Experts would choose their windows, volume, and price. Those operating terms remain a proposal until roster agreements and live booking are activated.
Who gets to teach changes
This is the part we care about most. Some strong operators will never record a course — they have day jobs, and a course is a production. A few time-boxed diagnoses inside their specialty could be more sustainable. That is the recruiting thesis behind the founding roster: build toward a network of people who do the work for a living, then activate the service only when coverage is real.