Create Professional Dashboards With Claude Live Artifacts

Sign Up here | Advertise with Us

AI agents now read your docs almost as much as humans do.

Mintlify analyzed 790 million requests across its documentation platform. The finding: AI coding agents account for 45.3% of all traffic, nearly tied with traditional browsers at 45.8%.

Two tools are driving almost all of it:

• Claude Code: 25.2% of total traffic, more requests than Chrome on Windows

• Cursor: 18% of total traffic

• Together they account for 95.6% of all identified AI agent traffic

The rest of the field, OpenCode, Trae, ChatGPT, and NotebookLM, is showing up but nowhere close.

One caveat: OpenAI's Codex doesn't send an identifiable user-agent header, so the real agent percentage is likely even higher.

The takeaway for anyone maintaining developer docs: your documentation now serves two audiences. Structure and machine-readability matter as much as clarity for human readers.

Try Mintlify today!

Claude has a feature 90% of people haven't explored yet.

It's called Artifacts. And depending on what you hand it, it can build you a live, interactive data dashboard in just about 90 seconds.

We wanted to see how far it actually goes. So we grabbed a sample Profit & Loss dataset off Kaggle, dropped it in, and asked Claude to turn it into something that looks like it came out of a McKinsey engagement. This is how it went.

What Are Artifacts?

Responses in Claude web normally live in the chat and disappear into the scroll.

Artifacts are different. They render in a separate panel on the right — live, interactive, running in your browser mid-conversation. Not a screenshot of a dashboard. The dashboard itself. You click things. Filters work. Charts update.

The newer version of this, inside Claude's desktop workspace called Cowork, is called Live Artifacts. Regular artifacts are snapshots — build it, lose it when you close the tab. Live Artifacts stick around. The dashboard you built Monday is sitting in your sidebar Friday, refreshed and ready. You're not re-prompting. You're reopening.

How to setup in 5 simple steps?

The Experimentation

We took 11,086 transaction rows from FY2017 to FY2019. 15 product categories, 3 store channels, 100+ producers.This was a sample file we downloaded from Kaggle here’s the link.

One prompt: 

Build a McKinsey-grade interactive executive dashboard from a Kaggle dataset — rendered live inside a Claude artifact."
Context:
I have a .xlsm file downloaded from Kaggle. I want to use it to demonstrate that Claude Artifacts can function as live, fully interactive dashboards — not static charts, but a real data product.
What I want built:
A single-page interactive HTML dashboard that ingests all meaningful data points from the uploaded XLSM and visualises them through:

KPI summary cards at the top (totals, averages, key metrics — whatever the dataset contains)
At least 3–4 chart types (e.g. bar, line, donut/pie, scatter) using Chart.js or D3, depending on data shape
Filter controls — dropdowns or toggles to slice data by category, time period, or segment (whatever columns exist)
A clean data table with sortable columns at the bottom for drill-down

Design direction — McKinsey / Tier-1 Consulting aesthetic:

Color palette: deep navy (#0A1628) background, white text, sharp blue accents (#1B6FDE), muted grey for secondary elements
Font: IBM Plex Sans or Libre Baskerville for headers, clean sans for body
Layout: CSS Grid, asymmetric column weights, generous whitespace
Charts: minimal gridlines, no chart junk, clear axis labels, subtle drop shadows on cards
No rounded corners overkill — sharp, architectural feel
Small animated counter on KPI cards on load

Constraints:

All in a single HTML artifact (no external dependencies beyond CDN)
Chart.js via CDN for charts
Data should be hardcoded from the XLSM after extraction (since artifacts can't read files at runtime)
Must be fully interactive — filters actually change what's displayed

Claude read the file, identified every sheet, ran aggregations across all the data, then started building. It knew what it was working with before writing a line.

The .xlsm format didn't extract cleanly. Claude tried one method, hit a column offset error, switched to Python, fixed the row indexing, and kept going. No intervention from me.

What Came Out Actually blew our Mind!

Five number cards at the top that count up when the page loads — total revenue, margins, best channel, fastest-growing category.

Six charts running at the same time. A bar chart, a donut, a line graph, a scatter plot — all live, all on one page.

One filter changes everything. Pick a year, the whole dashboard updates. Pick a channel, every chart shifts.

And a table at the bottom where you can sort and dig into the actual numbers.

All rendered inside the Artifacts panel. I didn't leave Claude once.

The insights weren't pre-labelled in the file either. Claude derived them. Alcohol up 69.4% from FY17 to FY19. Convenience Stores went from third-largest channel to 57.6% of all revenue. Delicatessen had the highest gross margin at 32.8%.

Where This Gets Insane

The domains where Live Artifacts genuinely changes things:

Data analysis & visualisation — drop any structured file in, get a working visual analysis. No Python setup, no analyst.

Revenue operations — MRR, ARR, churn, expansion. One artifact in your sidebar, always current.

Sales pipeline — deal stages, close rates, pipeline value. Connect your sheet, open it Monday, it's already updated.

Marketing budget tracking — every channel, every spend, every return. Live. Not a spreadsheet you have to remember to update.

And if you're a solo founder running lean — no data team, no Tableau seat, no BI subscription — this is the most disproportionate thing you can get your hands on right now. Infrastructure that used to require a hire, for $20 a month. The math is embarrassing.
Check out more resources

📣The Bottom Line 

Three tools usually go into presenting data well — Excel for numbers, Python or Tableau for charts, a designer for layout. This collapses all of that into one conversation.

The Kaggle file was a stand-in. The real version is your own quarterly numbers, campaign data, unit economics. Drop it in, describe what you want, watch the panel build while you're still typing.

That's the part worth paying attention to. 🤖 
What's Up in AI Team