I didn't expect this one to land as hard as it has. When Jess and I were brainstorming verticals, advisorlink was in the middle of the list — "AI homeschool portfolio + advisor report generator, $19/mo Pro tier, $49/mo Co-op Plus." I didn't have strong feelings about it. I do now, because the first three users who tried it were deeply, specifically grateful in a way that made it clear we'd hit a real nerve.

Here's the problem advisorlink solves, and why I think it'll end up being one of the portfolio's hero products.

The Homeschool Portfolio Problem

Depending on the state, homeschool families submit some kind of periodic evidence that their kid is actually learning things. In Washington, it's a quarterly or annual review with a state-approved advisor. In Oregon, test scores. In Idaho, essentially nothing. Every state has its own rules.

Washington's model is representative of the harder cases. Four times a year (or once a year for annual review), a homeschooling parent has to meet with a state-approved advisor and produce a portfolio: a collection of student work samples, organized by subject, with a narrative describing what the student learned. The portfolio is then reviewed, signed off on, and archived.

Here's what that actually looks like in practice. Parent sits down at 10 PM the night before advisor review. Opens Google Docs. Starts pasting photos of the kid's handwritten math work, photos of their essay assignments, screenshots of projects. Writes a narrative paragraph for each subject trying to describe what's shown in the samples and what skills they demonstrate. Exports as PDF. Shows up to the appointment looking exhausted.

This process takes 3-8 hours per review. For a family with multiple homeschooled kids, that's a miserable weekend, four times a year.

What advisorlink Does

Three things:

Work sample management. Upload photos of handwritten assignments, PDFs of essays, scans of math sheets, screenshots of projects. Tag each sample by subject (Math, Language Arts, Science, Social Studies, Art, etc) and by quarter. Everything stays organized.

Claude-generated narratives. For each sample, Claude analyzes the image or document and writes a paragraph describing what skill it demonstrates. For the full portfolio, Claude writes an overall narrative describing the student's progress across the period, in the tone an advisor wants to read.

PDF export. One-click generation of a formatted PDF: cover page with student name and review period, per-subject sections with work samples and narrative, overall progress narrative. Looks like a portfolio, not like a hastily-assembled scrapbook.

What Makes This Work

The thing that makes advisorlink useful is that the Claude-generated narratives are good enough to use with minimal editing. Given a photo of a kid's division problem and a subject tag of "Math — Grade 4," Claude reliably produces something like:

This sample demonstrates Ava's progression with long division, working through a 3-digit dividend with a 1-digit divisor. She's shown her work clearly, including the subtraction steps and the handling of remainders. Her presentation is neat and the arithmetic is accurate. This is consistent with grade-level expectations and shows the procedural fluency we'd expect at this stage.

That's the kind of sentence a homeschool parent would take 10 minutes to write from scratch. Claude takes about 3 seconds. Multiply that across 30-50 samples in a quarterly portfolio, and you're talking about turning an 8-hour task into a 1-hour task (mostly editing and curating).

The parent is still in charge. The parent decides which samples to include, edits the Claude drafts, and signs off on the final portfolio. Claude is doing the first-draft writing pass that was the bulk of the miserable work.

Privacy Is Non-Negotiable

This was the first conversation I had with every beta tester. Their kid's essays, their kid's math homework, their kid's family photos from project days — they do not want that data training somebody's LLM. They do not want a "cloud education platform" holding it. They do not want a vendor acquisition to mean their kid's first-grade writing samples end up on an S3 bucket owned by a company they've never heard of.

advisorlink is built the same way as every other Dangercorn vertical: Flask + SQLite, self-host friendly, your kid's work stays on your laptop. Claude is called through the API but the conversations aren't retained for training (per Anthropic's API terms). If you self-host, you can air-gap the whole thing and just not use the AI features when you don't want to.

The families I've talked to care about this deeply, and it was the single biggest "will you trust this product" factor. Not the UI, not the features, not the price — the data handling.

Build and Stack

Same template as every vertical: Flask + SQLite + dcst. About 1,800 lines of custom Python on top, mostly because the PDF generation pipeline is non-trivial and there's real UX work in the sample upload flow (drag-and-drop, bulk tagging, per-quarter organization).

Quickstart:

git clone https://github.com/Dangercorn-Enterprises/advisorlink.git
cd advisorlink
pip install -r requirements.txt
echo 'ANTHROPIC_API_KEY=sk-ant-...' > .env
python app.py

Pricing

What This Is Competing With

Essentially nothing built for the portfolio use case. There are homeschool curriculum platforms. There are online record-keeping tools (most abandoned or minimally maintained). There are state-compliance workflow tools for umbrella schools. None of them do what advisorlink does: take the work samples you already have and produce the formal narrative portfolio the review process actually requires.

The closest thing is a collection of Google Doc templates floating around homeschool forums. Those templates are static — they don't generate narrative, they don't organize by subject, they don't export to the format advisors want. They're a starting point, not a solution.

The moment a family realizes this turns an eight-hour task into a one-hour task, they're willing to pay $19 a month. That's a ten-minute sales conversation, not a three-month enterprise pipeline.

What's Next

Features on the near-term roadmap:

The state-specific templates is probably the highest-value work — once advisorlink knows the format Washington's Office of Superintendent of Public Instruction wants, every Washington homeschool family gets a lot of friction removed at once.

If You Homeschool

Pull the repo. Run it on your laptop. Upload a couple of your kid's work samples and see if the narratives are usable. Email me with what's wrong. This is the kind of product where the first 50 users shape the next 500.

And if your state isn't represented in the templates yet, that's 100% a fixable thing — tell me what format your advisor wants and I'll add it.