Analyze any data on screen and get the actual takeaways
A revenue dashboard, an A/B test result, a CSV in a spreadsheet, a customer table. Crade reads what is on screen and tells you what is going on, what to act on, and what is just noise.
| Category | Q3 | Q4 | Change |
|---|---|---|---|
| Revenue | $840K | $1.03M | +23% |
| Costs | $520K | $681K | +31% |
| Engineering | $280K | $310K | +11% |
| Marketing | $150K | $222K | +48% |
| Operations | $90K | $149K | +66% |
| Profit | $320K | $349K | +9% |
• Revenue growth strong at 23%, driven by enterprise deals
• Marketing spend increase (+48%) outpacing revenue growth
• Operations costs rising due to new infrastructure investments
• Net profit margin declined from 38% to 34%
Analyze this data for me
Key findings: • Revenue +23%, but costs +31% → margin shrinking • Marketing spend +48% with low conversion • Operations +66% from new infra → Recommendation: audit marketing ROI before next quarter
Most data on a screen is unread. A dashboard you glance at once a week. A report you bookmark and forget. A spreadsheet you scroll through hoping to spot the pattern. The bottleneck is not the data. It is the time to make sense of it. Crade reads the data on your screen and gives you the takeaways: what is going up, what is going down, what the cause is likely to be, and what to do about it.
What you put on your screen
- A dashboard with charts and metrics (Mixpanel, Amplitude, Plausible, etc.)
- A spreadsheet with raw data
- An A/B test result page
- A financial report or P&L
- A customer feedback CSV or NPS survey export
What you say to Crade
Or focused: "What is driving the drop in retention?", "Is the A/B test result significant?", "What should I act on this week?". Mention what you already know to skip the obvious.
Step-by-step: how to do this in Crade
Open the data view
Dashboard, spreadsheet, report. Make sure the key metrics or visualisations are on screen.
Click the Crade icon
Expand Crade. Floats over the data.
Ask for the analysis
"What is happening here?", "Where is the problem?", "What is interesting?". Crade reads the visible data and flags patterns.
Drill into specific findings
"Why did revenue drop in week 3?", "Is the marketing spend correlated?". Crade looks at the data and tries to reason about cause.
Ask for the action
"What should I do about this?". Crade suggests concrete next steps based on the patterns it sees.
Save the analysis
Copy the takeaways into your team's Notion, a Slack message, the meeting prep doc. The fast-write version of a real analysis.
What you get back
Structured takeaways: key finding, possible cause, suggested action. Crade is honest about uncertainty. For genuine causal claims it says "this is correlated" rather than "this caused that". For action recommendations, it ties them to specific data points.
Tips for better analysis
- Tell Crade your role and what you are trying to decide. "I am the product manager, deciding whether to ship the new onboarding" gives Crade the goal to tie findings to.
- Ask Crade to flag what is missing. "What would help me understand this better?" surfaces blind spots in the data.
- Verify any specific number Crade quotes against the source. AI is good at narratives, occasionally fuzzy on exact figures.
- Use Crade for hypothesis generation, not for the final analysis. "What are three possible causes?" beats "What is the cause?". The second answer might be wrong but presented confidently.
- For statistical significance, ask Crade to walk through the math: "Is this difference statistically significant given the sample sizes?". The reasoning is more trustworthy than the verdict alone.
Free vs Pro vs Premium
- Free ($0): reads visible data and gives narrative analysis. Plenty for occasional dashboard checks.
- Pro ($7.99/mo or $49.99/yr): higher daily usage, Agent mode (Crade can run SQL queries or scripts to dig deeper). Right tier for analysts and operators.
- Premium ($19.99/mo or $149.99/yr): 10x more daily usage. Right tier for data teams.
Frequently asked questions
Can Crade do statistical tests?
Crade can describe and apply common statistical reasoning (significance tests, confidence intervals, correlation vs causation), but it is not a stats package. For peer-review level analysis, use proper tools (R, Python, JMP) and ask Crade to interpret the output.
Does Crade read interactive dashboards correctly?
Crade reads what is visible. The static state of the dashboard. For interactive elements (hover tooltips, dynamic filters), the values are read at the moment of the prompt, not as you click around.
Will Crade make up numbers?
It tries not to. If a value is unclear or partially obscured, Crade flags it as "approximately" or asks you to verify. For exact precision (financial reporting, scientific data), always spot-check.
Can Crade analyse my customer data?
Yes, but consider privacy. The screenshot goes to the AI provider. If the data contains PII (real customer names, emails, payment info), either obscure it first or use Crade only with anonymised/sample data.
What about predictive analysis?
Crade can describe trends and project from them ("if this rate continues, revenue hits $X by Q4"), but it does not do real forecasting models. For ML-based forecasting, use a proper tool and feed the output to Crade for interpretation.
The whole loop in one sentence
Data on screen, one prompt, takeaways back. From staring at a dashboard for ten minutes to knowing what to do about it in thirty seconds.
A screenshot of a slide. A photo of a printed report. A frame from a YouTube video. Crade reads the table on your screen and gives you the rows in CSV format, ready to paste into a spreadsheet.
Three plans, three products, three job offers, three Airbnbs. Open them in separate tabs or windows. Crade reads them all and tells you which one fits your situation best, and why.
A 30-page vendor agreement, a 50-page report, a research paper. Crade reads what is on your screen and gives you the structured summary: what it is about, the key points, what to act on.