The Grain platform

One tag. The whole funnel, observed.

Drop one script and you get the analyst, the investigation, and the evidence — not eight tools to wire together. Every number on this page is something a visitor can click straight through to. No data team. No SQL.

14-day free trial · no credit card · cookieless by default

The loop that finds the leak

Every product below is a step in one loop: capture behavior, ask a question, follow it to the cause, and land on the exact session that proves it.

01

Drop the tag

One script, server-side ready. Web, WordPress, iOS, and Android.

02

Ask Kai

Plain-language questions over your live traffic and sessions.

03

Investigate

Kai works the funnel backward until the cause has a name.

04

Cite the session

Every answer links to the exact replay, heatmap, or row behind it.

The AI analyst

Ask Kai anything. Get the number and the chart.

Kai reads your traffic, sessions, and goals the moment they land. Ask in plain language and it answers with the figure, the breakdown, and a link to the evidence — not a dashboard you still have to read.

  • Plain-language questions over live behavior data
  • Answers ship with the chart and the citation attached
  • Bring your own key (OpenAI, Gemini) on private deployments
Ask Kai in the demo~3s to first answer
Kailive
You

Which traffic source has the best signup rate this month?

Kai

Organic search converts best at 6.2%, well ahead of paid social at 1.4%. It’s 38% of traffic but 61% of signups.

Open the chart
Deep Investigation

Point Kai at a drop. It works the funnel backward.

A six-phase audit that doesn't stop at 'conversions fell 14%.' Kai decomposes the metric, segments the affected users, clusters the sessions, and names the cause — with the proof bundled into a shareable report.

  • Six-phase audit: decompose → segment → cluster → cause
  • Statistical context and p-values, not vibes
  • One-click shareable report for the team
See an investigationFound a $14K/mo bug in 8 minutes
Deep Investigationcheckout · −14%
  • 01Decompose the metric
  • 02Segment affected users
  • 03Cluster the sessions
  • 04Correlate with releases
  • 05Name the causerunning
  • 06Write the reportqueued

Card step breaks on mobile Safari after the May 9 release. 712 sessions affected.

Replay & heatmaps

Every finding is citable. Jump straight to the session.

A number is only useful if you can see it happen. Click any metric and land in the exact sessions behind it — full replay, click maps, and scroll maps, cookieless and EU-hosted by default.

  • Pixel-accurate session replay (rrweb)
  • Click, scroll, and movement heatmaps per page
  • Cookieless capture — no consent banner required
Open a replayEU-hosted · cookieless
Sessions on /checkout712 abandoned
  • mobile SafariiPhone 150:42
  • mobile SafariiPhone 141:08
  • desktop ChromemacOS2:15
  • mobile SafariiPhone 130:55
Click density · card field94%

The rest of the stack

Same tag, same data. Each of these reads from the one event stream Kai already understands.

Privacy is a feature, not a disclaimer

EU-hosted

Data stays on European infrastructure. GDPR-ready out of the box, no processor gymnastics.

Cookieless by default

Recall identifies returning visitors without cookies or fingerprinting — no consent banner needed.

Bring your own key

On private deployments, point Kai at your own OpenAI or Gemini key. Your prompts, your model.

Install once. See everything.

One tag turns on the whole platform. Ask your first question before your coffee’s cold.