Explain the role of templates and presets in reducing the blank-page problem.

Kestwick & Voss Appraisals equips its appraisers with FormFast after every site inspection. Selene Marchetti has appraised homes for eight years and files two to three reports a week, usually the same evening she inspects the property.

How to answer this, stage by stageSix moves, in the order you'd actually say them.

Let's learn

FormFast reads an appraiser's inspection notes and, once it detects the property type, prefills a full report skeleton, standard sections, boilerplate disclosures, typical measurements, ready to review in seconds.

Before it, Selene opened a blank report form after every single inspection and rebuilt the same structure from memory each time, roughly forty minutes before she'd written a single finding specific to the actual property.

The first version of FormFast picked a preset silently, based on county tax records, with no way to see or change it before the report opened for editing.

Three of these four steps have nothing to do with the actual property. A preset exists to erase exactly those three.

And here's the turn: the blank-page problem was never about Selene lacking things to say. It was about being asked to invent the report's whole shape before she ever got to say any of it.

Every callout here does one job: handle the structure, and stay out of the one section that's actually hers.

At its worst: a property quietly converted from single-family to a duplex still shows as single-family in county records. FormFast's silent preset carries single-unit occupancy language straight into the draft, and it slips past review because nobody had any reason to double-check a section that reads like boilerplate.

What I would leave alone: for a standard, recently-built subdivision home with clean county records, the auto-selected preset is right almost every time, and adding extra confirmation ceremony there would only slow Selene down for nothing.

The lesson: a preset earns its place by solving exactly one problem, deciding the shape of the page, and it loses trust the moment it quietly tries to solve a second one too.

Now here is the same thing as a storyThe short version is above. Read this for how the anchor actually got picked.

Selene can walk a property once and already know which three comparable sales she'll cite before she's back in her car. Eight years of appraising built that instinct.

Writing the report itself was always the slow part, not because she didn't know what to say, but because every report started from the same blank shape she had to rebuild from memory first.

The anchor only has to prove itself on the right side. That's the entire point of designing for it in advance.

FormFast's first version picked a preset the instant she opened a new report, single-family, based on the county record, with no indicator of what it had chosen or why.

The third branch, asking first when records are unclear, didn't exist in the version that nearly shipped a wrong report.

On a property at the edge of town, county records still listed it as a single-family home, three years after the owner had converted the attached garage into a rented second unit. Selene's own notes mentioned a second unit, but the preset's boilerplate, already filled in before she'd typed a word, still described one household, one entrance, no separate utilities.

None of these three ever got automated, in either version. The fix was never about doing less writing. It was about seeing the boilerplate before trusting it.

She caught the mismatch only because she happened to reread the boilerplate section closely that week, after a colleague's report had been kicked back for a similar issue.

With the redesigned first screen, the chosen preset now shows plainly before she starts, "Single-family preset selected," with a one-tap way to change it, and the moment her own notes mention a second unit or separate utilities, a flag appears: "This preset assumes single-unit occupancy. Your notes mention a second unit. Recheck?"

The same converted-duplex property now gets caught in the first two minutes of drafting, not discovered by luck a week later after a colleague's report already got kicked back.

I would take back the silent auto-selected preset. It made the first-open experience feel effortless, and that felt like a win at the time. It took one appraiser catching a real mismatch by luck to see that invisible and effortless aren't the same thing when the boilerplate can end up in a legal document.

SPARK, the preset as the anchorFive letters. The third one is the whole design decision.

The recap, one line per letter: situation is the forty minutes lost rebuilding structure each time, payoff is trading that for time spent on real findings, anchor is the editable preset with boilerplate prefilled and findings left blank, risk is a stale county record misclassifying a converted property, and keep out is the narrative, judgment, and conclusion staying fully manual.

And if you want to be sure it really works, try it somewhere elseSame five letters, a textile mill instead of an appraisal office.

Hearthloom Textiles uses an AI tool that drafts defect reports for a quality line, prefilling a report template based on the defect category a camera flags. Farida Renshaw leads quality control and reviews flagged rolls of fabric before they ship.

Mapped onto SPARK: situation is Farida writing a defect report from a blank template for every flagged roll, describing severity and likely cause from scratch each time. Payoff is trading that blank-template habit for reviewing a prefilled defect category and writing only the roll-specific detail. Anchor is a defect-category preset, weave slub, dye migration, and so on, that fills in standard severity language and likely-cause boilerplate, while leaving the specific measurement and disposition decision blank. Risk: a genuinely novel defect, a new blend nobody's flagged before, gets misclassified into the nearest known category, carrying wrong boilerplate about cause. Keep out: the actual pass, hold, or reject decision on a roll never gets automated, staying entirely Farida's call.

The bottom-left corner, rare and low-confidence, is exactly where a preset has to say "not sure" instead of guessing.

Swap the trigger and it still runs.
Speed: an interviewer caps you at sixty seconds. Say "presets solve the structure problem, never the judgment problem," and stop.
Cost: no engineering time to build mismatch detection this quarter. Say so honestly, and start by simply always showing which preset got picked, since visibility alone catches most mismatches without any detection logic at all.
The model gets better, for real: if property-record data quality improves overall, a rare misclassification becomes more dangerous, not less, since nobody's expecting to double-check a preset that's almost always right.

Where people run it wrong.
They let a preset creep into filling the judgment sections too, since "it's already drafting the boilerplate, why not the findings," which quietly removes the one part of the report that had to stay human.
They auto-select the preset silently to make onboarding feel seamless, which is exactly what let a wrong assumption travel furthest before anyone caught it.
They treat "reduce the blank-page problem" as a writing-speed problem, when it's really a structure-versus-judgment problem with two very different fixes.

How to use it live. When someone asks about templates and presets, ask yourself one question first: which part of this blank page is structure, and which part is judgment. Say the split out loud before saying anything about automation.

Flashcards (tap any card to flip it)

Check yourself Score: 0 / 0