Skip to main content

Ask it what happened.

Type a question in plain language. The answer comes back with every figure it used, cited to the source, and a chart when one helps. If you like it, one click makes it a dashboard or a deck.

Live in early access. Answers read from the store, so they cover whatever your connected sources hold.

Every figure cited and checked.

The model can only choose from figures already held in the store, which plain code pulled from your sources. Before you see an answer, every number in it is checked against those figures. A number the tools did not supply fails the whole answer rather than slipping through.

  • Figures appear as chips naming the source and the metric.
  • Follow any chip back to the exact request and response that produced it.
  • If a question cannot be answered from held data, it says so, and says why.
Ask answering how organic search did last month: 40,583 sessions and £50,548 in revenue for September, both from Google Analytics, with a daily chart and the two figures it stands on listed underneath
A real answer on a demo workspace, with the figures it stands on listed underneath.

Questions that cross sources.

Because the figures are kept together, a question can put traffic beside revenue or pipeline beside spend. No single provider does that join for you.

  • Google Analytics 4, Search Console, HubSpot, Mailchimp and Stripe are live today; the rest are in progress.
  • Each source adds its metrics and dimensions to what Ask can reach.
Ask answering how the Shopify store did over 28 days: total sales of 202,121 across 2,244 orders, an average order value of about 90, and the online store taking 78 percent, with every figure listed underneath and offers to turn it into a report or a dashboard
Shopify sales by channel, and the offer to turn the answer into a report or a dashboard.

If it isn't held yet, it gets fetched.

If the store does not hold a figure and the source can supply it, Confluxly fetches it first and then answers. If the source cannot supply it, because the account did not exist then or the scope was never granted, the answer says that plainly instead of showing a zero.

Explorations, for when you want the table.

Pick a measure, split it by any dimension the source offers, and compare two periods. The arithmetic is done by code, never by the model, so a rate is weighted properly and a unique count is never added twice.

Explorations: total sales broken down by channel over the last 28 days, shown as bars with a table beneath: online store 157,636, Shop app 16,230, point of sale 11,955, Instagram 10,186, draft orders 6,115
Pick a measure, a breakdown, a range and a shape. Code does the arithmetic.

Then the rest of your stack.

Planned

The plan is to add the tools that hold the rest of the day, such as the inbox, the calendar and the support desk, with you choosing what it may read. Today Confluxly reads figures only.

What it does not do yet.

Written down so you do not have to ask. When one of these changes, the changelog says so.

Read the changelog
  • It cannot tell you why a number changed. Causal insight, anomaly alerts and a daily digest are planned, not built.
  • It reads figures only. Documents, inboxes and calendars are not connected, and the sensitivity levels for them are a plan.
  • Google Analytics 4, Search Console, HubSpot, Mailchimp and Stripe are live today. Other sources are in progress and marked as such on the sources page.
  • Answers are shared with your workspace, not with a public link.

Questions people ask about this.

The same store, other outputs.

See it on your own numbers.

Get early access and we will build the first draft from one of your own sources.

Get early access

Where your data converges.

Get early access