Evarist reads behavioral analytics so your AI can reason about your traffic. This page describes, precisely, what we collect, where it lives, and the boundaries that protect it — written for you and for your security reviewer. Every claim reflects how the product is actually built.
The full inventory — and the things the tracker is built to never touch.
The tracker collects behavioral events from your site: pages viewed and their titles, referrers and traffic source, browser timezone, scroll depth, visible time on page, plus the interactions and conversions you choose to tag.
Identifiers are two first-party cookies holding random IDs — a visitor ID (13 months) and a session ID (30 minutes). Both are generated in the browser as random UUIDs, derived from nothing about the device. Country, browser and device family are derived on our servers at ingestion from standard request headers; the visitor's IP address is masked before storage and the full IP is never retained.
If you install our Shopify app, we also read a narrow set of order facts from Shopify's Admin API — order number, timestamps, currency, status, and the monetary totals (net payment, price, refunds, discounts) — plus your product and inventory records. That is what lets Evarist tell you which traffic sources actually produced revenue. We never ask Shopify for customer names, emails, phone numbers or addresses: the app holds no customer-data permissions, and the queries we send contain no customer fields. Individual orders are deleted after 65 days; only daily totals per traffic source are kept beyond that.
What we never collect:
Names, emails and phone numbers collected by Wisepops campaigns live in a separate system Evarist cannot query.
One customer's data must never surface in another's AI answers — so we enforce it at the lowest layer available.
SELECT — a crafted query, UNION, or subquery cannot escape it.A chatbot sits on top of this data, so the AI layer gets its own boundaries.
Not from the prompt. Your AI's requests carry your token; the token selects your scoped database account; nothing the model writes can widen it. A prompt-injection attempt can phrase any query it likes — the engine still only returns your rows.
Database errors are sanitized before the model sees them; raw internals are logged only on our side. Queries are rate- and size-capped per user.
Server-side, Evarist calls one model provider: Anthropic. Three uses — profiling your public website pages, suggesting event tags, and generating analytics insights from query results. Visitor PII is never sent. Neither Evarist nor Anthropic trains models on this data.
When you connect your own AI client (claude.ai, ChatGPT…), that provider processes your questions under your agreement with them — it is your processor, not our sub-processor. Our sub-processor list spells out the distinction.
All traffic is encrypted in transit (TLS): the tracker, the MCP connection, the web app. Account and configuration data is stored on encrypted volumes (AWS EBS), and its backups are AES-256 server-side-encrypted in S3.
On AWS (us-east-1, United States) — the analytics infrastructure Evarist shares with Wisepops. Our sub-processors — AWS, Cloudflare, Postmark, Anthropic — are listed publicly with their purpose and location at app.evarist.ai/subprocessors.
For EU/UK customers: transfers rely on the EU-US Data Privacy Framework and/or Standard Contractual Clauses with the UK Addendum, as set out in the DPA.
With the documents to forward.
We would rather list these plainly than imply them. This section shrinks as we grow.
Questions your reviewer wants answered directly? support@evarist.ai