01
What happened
The number, the trend, the deltas — not a tile to interpret, an answer to read.
Data Apps · The Operating Loop
Analytics has four invariant questions. Most BI tools answer the first and leave you alone with the rest. Spotonix answers all four — for every signal you watch, automatically and honestly.
The four questions
Dashboards render the chart and walk away. Text-to-SQL agents return a number and forget the question. Neither tells an operator what to do — or remembers what was decided last time. The job is to answer all four, for every signal you own, automatically.
01
What happened
The number, the trend, the deltas — not a tile to interpret, an answer to read.
02
Why
The driver, gated by process-control so normal noise doesn’t get a story. “Why unknown — investigate” when the data won’t support a claim.
03
What’s next
The trajectory if pace holds. The forecast a human analyst would make — with the same caveats.
04
What to do
One trustworthy recommendation. The play that worked before, or the next step to investigate. Never three charts and good luck.
I used to walk into my Monday review hoping nothing had blown up over the weekend. Now the review is already written — what changed, what we decided last week, and whether it actually worked. I spend the meeting deciding, not pulling numbers. I haven’t been blindsided since.
— Rohit Laungani, Enterprise Architect
The operating loop
You never learn a new tool. You do the verb you were already going to do, and the product notices the pattern and offers the next step. It’s the same card — gaining capability as your own work earns it.
Day 1 · Ask
You ask a normal business question. A dashboard would hand you a chart and walk away. Spotonix hands you the chart and the four answers an operator actually needs — composed into one card.
WhatTotal Sales fell 8% to $42.0M
WoW −1.2% MoM −3.4% YoY −8.0%
WhyReturns are the driver to watch
Next↘ likely to keep slipping if pace holds
DoInvestigate driver →★ Watch
Day 7 · Watch & Brief
Tap ★ on the one number you care about. From then on, the system carries it for you. By the end of the week, your morning Brief is a ranked portfolio of what changed and why — with an honest “all quiet” when nothing did.
The watch atom — what gets carried forward
A watched signal is a semantic object — not a saved chart. That’s why Spotonix can keep watching it: it knows what the answer means.
Monday · Morning Brief 6 watched · refreshed 6:00am
●Total Sales −8% off-track · Why: Returns ↑
●Total Returns +12% over guardrail
●Web Sales −15% real signal (out of normal range)
●Order Count +4% on-track
○Net Profit flat · Why unknown — investigate →
── All others quiet — steady, nothing to escalate ──
Day 14 · Review & Trails
Three related signals become a Board. A Board with a rhythm becomes your weekly Review — the agenda, the carryover, and last week’s decision already assembled. Every investigation becomes a Trail, so the memory of why is carried in, not re-argued.
Agenda
• Total Sales down 8% — Returns the confirmed driver
• Web Sales out of normal range
Carryover from Week 17
• “Discount slow-moving Electronics bundles” — measuring
Decision (recorded)
▸ Owner: Mike · Expected: Sales recovers next refresh
From the Trail
● Fact — Decline concentrated in Electronics, web
● Driver — Returns → Sales ✓ confirmed
● Decision — “Discount Electronics bundles” (Week 17)
● Outcome — Week 18: Sales +2.1% — moving toward plan
Day 21 · Apps
When a play has worked before, the card offers to apply it again — through your existing automation, at a trust level you climb deliberately. Spotonix is the brain, not the hands: it proposes, you approve, and the outcome returns to next week’s Brief.
Trust ladder · Total Returns
Live today: Watch · Notify · Propose · Roadmap: Approve · Auto + undo · Autopilot
Playbook proposal · “Surcharge pass-through”
Worked Week 14 → returns 9.2% → 6.8% in 11 days
Owner: Mike · Expected: ≤7.5% in 2 weeks
✓ Approve & Run✎ Edit× Skip
v1 acts only through safe, reversible channels — Slack, ticket, writeback, webhook. Your existing automation does the transactional work. Every rung is earned by measured outcomes, with an undo window. Down is one click.
Why this becomes a habit
Dashboards aren’t sticky because they’re passive. Text-to-SQL isn’t sticky because it’s transactional. The morning brief is sticky because it’s a daily ritual with variable reward — and an investment loop that makes tomorrow’s brief more useful than today’s.
Trigger
Morning notification.
Same time, every day. Slack, email, or app — your choice. The cue makes it a ritual.
Action
Scan in 30 seconds.
Five minutes if there are attention items. Mark expected. Snooze. Investigate.
Reward
Variable — sometimes “holy sh*t.”
Boring quiet days. And the days the system catches what no human would have.
Investment
Every click teaches the system.
Mark expected, confirm a driver, change a guardrail — tomorrow’s brief is more relevant. The habit reinforces itself.
Why this can be trusted
Spotonix is built to be trusted with judgment, which means it has to be honest about its limits. Four mechanisms keep it that way — and one ladder keeps you in control.
01
Conformed measures
One “Sales” reconciled across store, catalog, and web. The number means the same thing everywhere — or the system tells you it doesn’t.
02
A readable plan
Named nodes and edges you can check in ten seconds. Compiled to SQL only after the algebra closes — not a black-box prompt.
03
xMR process-control
Statistical gate on causal claims. Normal variation gets no story. Real signal — outside the natural range — gets escalated.
04
Honest unknown states
“Why unknown — investigate” instead of a fabricated driver. A quiet day says “All quiet,” not a blank screen.
And one trust ladder you climb deliberately:
One rung at a time. Always at your command. Moving down is one click.
Common questions
The questions every prospect asks in the first thirty minutes. Answered here so the eval starts ahead.
A dashboard renders; Spotonix decides. A dashboard is a grid of tiles you have to interpret every time you open it — Spotonix's Brief is judgment. It tells you what changed, why, what's next, and what to do, and it remembers the decisions you made and whether they worked. Dashboards forget; Spotonix runs a loop.
Chat answers once and forgets. There's no monitoring, no decisions, no outcome loop — and nothing stopping a confident-but-wrong answer. Spotonix grounds every answer on your team's conformed measures and a readable plan you can check before any SQL runs, gates causal claims with process-control, and persists what it learns so the hundredth question is faster and sharper than the first.
Four things. (1) Conformed measures — one “Sales” reconciled across store, catalog, and web, so the number means the same thing everywhere. (2) A readable interpretation plan — named nodes and edges you can read in ten seconds — that compiles to SQL only after the algebra closes. (3) xMR process-control gates the “Why”: normal variation gets no causal story. (4) Honest unknown states — “Why unknown — investigate” instead of a fabricated driver.
Spotonix studies your existing artifacts (schema, dbt models, BI assets, past queries) to build a Context Graph that is yours alone; your data is not used to train shared models. Specific deployment, residency, and access controls are covered as part of beta onboarding.
Spotonix is the brain, not the hands. The v1 action surface is deliberately bounded and reversible: post to Slack or email, open a ticket, append to a curated writeback table, fire a webhook, record a decision. It does not mutate your CRM, order, or inventory systems directly — your existing automation (Workato, Zapier, n8n, dbt) does the transactional work; Spotonix triggers it. Every action runs at a trust level you set, and moving up the ladder is always the system's suggestion based on measured outcomes, never a default. Moving down is one click.
A data warehouse, and ideally the dbt models and BI assets your team has already built — Spotonix learns your business from that corpus before the first question. The more of your team's prior work it can study, the better the first answer.
It's in private beta. Request access and we'll set up a guided evaluation against your own data. Pricing is handled per design-partner engagement during beta.
Beta
Data Apps is in private beta. Bring your own warehouse and metrics; we’ll spin up a guided evaluation against your data.