Buyer’s guide
Written by a clinical psychologist who signs reports, and who also builds one of these products. Everything below is a question you can put to any vendor, including us.
Three different purchases get sold with the same demo. Typing speed on documentation you already know how to write. A draft that arrives in a shape close to your template. Or a system that holds the case and reasons across it.
All three are legitimate. They are priced differently, they fail differently, and they leave you holding different amounts of the work. Sorting the market into three kinds before you look at features saves a lot of demo time, and the field guide places every tool we know of into one.
Before a feature list, ask for these two things live, on a case with something messy in it. Both take about ninety seconds and tell you more than an hour of slides.
One. “Show me the evidence that contradicts this diagnosis.” Not a paragraph that sounds careful. The T-scores, the instrument, and the session the divergence came from. A system that cannot produce those is not reasoning about the case; it is writing around it.
Two. “Show me where this claim came from.” Pick a sentence in the draft at random and ask to open its source: the measure, the document, the session note. A fluent restatement that points nowhere is the thing that gets found later.
We published these because nobody else had, and they are written to be applied to any platform, ours included. As questions for a sales call:
1. Provenance
Every clinical claim resolves to an identifiable source.
2. Disconfirmation
The evidence that argues against a conclusion is surfaced, not only the evidence that supports it.
3. Approval
Nothing machine-generated enters a record without a logged clinician approval.
4. Voice
The report reads in your voice and follows your standards, because you are the one who signs it.
5. Auditability
Every action carries a timestamp, an actor and a role.
6. Disclosure
AI use is stated in the record itself, not buried in a license agreement.
Worth saying plainly, because it cuts against our own interest as much as anyone else’s: a very low price and an instant free trial are information. Not a red flag. Information.
Reading a whole battery, holding structured data, carrying provenance on every claim and running reasoning across the full file is expensive per report. It also needs configuring to your measures, your templates, your voice and your workflow before it produces anything you would sign. That work happens before you get value, which is why this end of the market demos rather than self-serves.
A tool you can be inside of in ninety seconds with a card is, almost by definition, doing something lighter: a block editor with good phrasing, or a scribe. Those are real products that solve real problems, and if that is what you need, the speed is a feature, not a compromise. The mistake is buying one while believing you bought the other.
During any trial, do not run the clean case. Run the one where the parent and teacher disagree, the one with the invalid profile, the one where your impression from the room does not match what came back in the post. Then check four things in the draft:
Then read the legal, ethical and workflow guardrails and check the process around the tool, which is where most practices actually get caught out.
See every AI report writer for psychologists, sorted by kind →