Alpha — early access. This page describes an early-stage product; features and pricing may change.
PraxisKB Practice knowledge base

Your practice's knowledge, actually findable.

We build and curate an AI-searchable knowledge base from your practice's own documents — SOPs, protocols, insurance workflows, the stuff that lives in one person's head. Cited answers, not a chatbot guessing. HIPAA-aware from day one.

Who it's for

Independent dental & specialty-medical practices, first.

The initial focus is independent (non-DSO) dental practices and small specialty-medical clinics in San Diego County — expanding to small law firms once the healthcare pilot validates.

The problem

When someone leaves, the knowledge leaves with them.

Independent practices run on institutional knowledge that mostly lives in one person's head or a scattered mess of PDFs, binders, and old emails — not written down anywhere searchable. Industry writeups on tribal-knowledge risk describe exactly this pattern: "if a high-level person in the firm retires or leaves without imparting the wisdom of what they know... you have a gaping void."

Onboarding

Onboarding takes weeks

New hires learn "how we do things here" by asking around, not by reading anything.

Exceptions

Nobody remembers the exception

"How did we handle this insurance denial last time?" — answered from memory, if at all.

Turnover

One departure, one crisis

The office manager who knew everything leaves, and three months of scramble follows.

Pain pattern corroborated by two independent secondary/consultant sources (professionalstrategygroup.com; personiv.com) — no raw first-person operator quote has been captured yet for this specific vertical. See RESEARCH.md §4.

What it is

The KB build-out, done for you — not another seat to configure.

Most "AI knowledge base" tools sell you software and leave the setup to you. We do the setup: intake your documents, extract and structure them, and hand back a knowledge base your staff can actually ask questions of.

Step 1

SOPs, manuals, insurance playbooks, onboarding docs — scanned or digital, sent as-is.

1. Intake

Send us what you've got — SOPs, manuals, insurance playbooks, onboarding docs, scanned or digital. We handle the mess.

Step 2

Extract → tag → structure — a knowledge base, not a document dump.

2. Build & curate

Documents get extracted, tagged, and structured into a knowledge base — not just dumped into a search box.

Step 3

Plain-English question in, cited source document out.

3. Cited answers

Staff ask a question in plain English and get an answer with the source document shown — verify before acting, don't guess.

Compliance

No patient document is touched before a signed BAA.

This isn't a footnote — it gates the whole delivery pipeline. Under 45 CFR §164.308(b)(1), a covered entity may only let a business associate handle PHI after obtaining "satisfactory assurances... that the business associate will appropriately safeguard the information." Building your KB makes us that business associate the moment we ingest anything containing PHI — so a BAA with your practice is executed first, and the AI vendor used for document extraction is confirmed BAA-covered before it touches anything, not assumed.

Landscape

What you're probably already paying for — and what it doesn't do.

Real current prices, not comparable on one axis (per-seat SaaS vs. enterprise contract floor vs. a services retainer) — so a table, not a chart.

VendorPriceWhat it actually gives you
Egnyte (compliant storage)$22–39/user/mo Where files live, HIPAA-eligible on the Enterprise Lite tier only. No AI, no search, no KB layer.
Tettra (generic wiki)$8/user/mo, 10-seat min A 5-person practice still pays for 10 seats. No compliance posture, no setup help.
Glean (enterprise AI search)$50–60K/yr min, 100+ seats (unverified) Structurally inaccessible below ~100 seats.
Practice Archive (accounting KB)No public price (beta) Same RAG idea, but accounting-only, self-serve software — not implemented for you, not healthcare.
PraxisKB$1,500–3,500 setup + $250–500/mo Someone builds and curates it from your own documents. Sits on top of storage you may already have.

Sources: tettra.com/pricing; egnyte.com/pricing; practicearchive.com/about (fetched 2026-07-29). Glean figure aggregator-corroborated, not directly fetched — see RESEARCH.md §3. PraxisKB pricing is a derived starting anchor, not yet tested against a real buyer — see STRATEGY.md §2.

Pricing

A setup project, then a light monthly retainer.

ItemPriceCovers
Setup $1,500–3,500
one-time
Scoped to document volume & complexity. Covers intake, extraction/tagging, KB structuring, and BAA execution.
Ongoing curation $250–500/mo
retainer
New documents ingested, KB pruned, quarterly compliance review of the BAA chain.

Where it pays for itself

Below is a hypothetical scenario, not a measurement — it exists to show the shape of the math, not to promise a result.

Illustrative time-saved-vs-dollar-value example Bar chart with three bars — 1 hour/week saved is about $150/month, 2.5 hours/week saved is about $375/month, 5 hours/week saved is about $750/month — against a dashed $400/month reference line for blended service cost. Axis runs from $0 to $800. ILLUSTRATIVE EXAMPLE — assumes $35/hr loaded staff cost × 4.33 weeks/mo $0 $200 $400 $600 $800 $400/mo — blended service cost 1 hr/wk saved ~$150/mo 2.5 hrs/wk saved ~$375/mo 5 hrs/wk saved ~$750/mo

A practice where even one staff member loses a half-hour a day to searching for or recreating documents clears this with margin. Assumptions are this pitch's own derived figures (STRATEGY.md §2), not an independently sourced study — Stage-4 flagged for a primary time-loss source if one exists.

Demo

Cited answers, not a chatbot guessing.

Preview below — fictional practice, sample content, not a live product. The full simulated demo has 3 scripted scenarios (front-desk insurance question, new-hire onboarding, and a compliance-flavored patient-record boundary check).

SIMULATED — ILLUSTRATIVE MOCKUP
Riverbend Family Dental — knowledge basedemo
"What's our process when an insurance claim comes back denied?"
Front desk checks the denial code against the payer's EOB first Front-Desk-SOP.pdf, p.4. If it's a coding issue, route to Dr. Alvarez for a corrected claim before resubmitting Billing-Playbook.pdf, p.11. Patient gets a status call within 2 business days Patient-Comm-Policy.pdf, p.2.

Every answer shows its source document — staff verify before acting, the tool doesn't replace clinical judgment.

Open the full simulated demo (3 scenarios) →

Objections

The questions we'd ask too.

Why not just use ChatGPT/Claude directly on our own files?

Doing that safely with patient documents requires a BAA with the AI vendor that a solo practice isn't set up to negotiate — plus the actual curation work (what's worth ingesting, how it's structured, keeping it current) that consumer AI tools don't do for you. The service is the labor and the compliance chain, not the model call.

How is this different from Egnyte, Practice Archive, or Glean?

Egnyte stores documents with no AI/search layer. Glean is priced for 100+-seat enterprises. Practice Archive is accounting-only, self-serve software with no implementation labor and no healthcare focus. This is the only option we've found that's a done-for-you build for a 5–25-person healthcare practice specifically.

Is our patient data actually safe / HIPAA compliant?

No PHI is ingested before a signed BAA with your practice, and the vendors used are contractually confirmed BAA-covered before any document touches them — that's a gating step in delivery, not a claim on a page.

What if the AI gives a wrong answer from our documents — who's liable?

The KB is built as cited-answer retrieval — every answer shows its source — with staff trained to verify before acting on anything patient-care-critical. It surfaces where an answer came from; it doesn't replace clinical judgment.

We're only 6 people — is this worth the money?

Run the breakeven math with your own numbers: at roughly $400/month blended cost against a $35–40/hr loaded staff rate, about 2.5–3 hours/week of no-longer-needed document searching pays for it — a small bar for most practices with any real tribal-knowledge pain.