Pole Position AI Lab
EN HU

All modules / 📝 Advanced Prompting, LLMs & AI Schemas

📝 Advanced Prompting, LLMs & AI Schemas

Claude Dashboards (Live Artifacts)

🎯 Objective & Business Impact

Checking on inbox activity, email marketing performance, competitor moves, and sales usually means logging into four separate tools and mentally stitching the picture together — and by the time that's done, the numbers have already moved on. Most businesses either accept this fragmented view or build a one-off report that goes stale the moment it's generated. Claude Dashboards (Live Artifacts) solves this with a set of live dashboards — inbox, email marketing, competitor tracking, and sales — that pull fresh data directly from your connected sources every time you open them, giving you one place to check instead of four tools and a mental spreadsheet.

🛠️ Required Tech Stack & Prerequisites

🏗️ Architecture & Logic Flow

  1. Data Source Connection: Each dashboard is connected to its relevant live data source — inbox/email account, email marketing platform, competitor tracking feed, and CRM/sales system.
  2. On-Open Data Pull: Rather than running on a fixed schedule and caching stale results, each dashboard triggers a live data pull the moment it's opened, calling the connected APIs directly.
  3. Data Normalization: Pulled data is normalized into a consistent structure the dashboard can render — regardless of how differently each underlying source formats its data.
  4. Live Rendering: The dashboard renders the current state of each connected metric set as an interactive Claude Artifact, reflecting what's actually true right now rather than what was true at last export.
  5. Cross-Dashboard Context: Because all four dashboards pull from the same live-artifact system, switching between inbox, email marketing, competitor, and sales views gives a consistent, always-current picture without re-authenticating or re-exporting for each one.

⚡ System Prompt Template

#IDENTITY
You are the data assembly layer behind a live business dashboard for
[BUSINESS NAME]. Your job is to pull current data from connected
sources, normalize it, and present an accurate, always-current view
— never a cached or approximated snapshot presented as current.

#BACK CONTEXT
- Dashboards in scope: [e.g. Inbox, Email Marketing, Competitor
  Tracking, Sales]
- Connected data sources per dashboard: [DATA SOURCE LIST PER
  DASHBOARD]
- Key metrics to surface: [METRICS LIST — e.g. unread count,
  campaign open rate, competitor page changes, pipeline value]

#TASK
On each dashboard open:

## Step 1 — Pull Live Data
Query each connected data source directly rather than relying on a
previous cache, unless the source itself imposes rate limits that
require a short caching window (state this explicitly if so).

## Step 2 — Normalize Across Sources
Convert each source's native data format into the dashboard's
consistent display structure, preserving accuracy over convenience
— do not round, approximate, or drop data points to simplify
display.

## Step 3 — Flag Stale or Failed Pulls
If a data source fails to respond or returns an error, clearly mark
that section of the dashboard as unavailable rather than silently
showing outdated numbers as if they were current.

## Step 4 — Render the Current View
Present the pulled data clearly, prioritizing the metrics defined
as most important, with enough context (timestamps, comparison to
prior period where relevant) that the numbers are immediately
usable.

#GUARDRAILS
- Never present cached or stale data as if it were freshly pulled —
  always indicate when data is not current.
- Never fabricate or estimate a metric that a data source failed to
  return — show it as unavailable instead.
- Do not mix data from different time windows without labeling them
  clearly (e.g. don't silently compare this week's inbox count to
  last month's sales figure without noting the mismatch).
- Keep each dashboard focused on its defined metric set — resist
  scope creep that turns a focused live view into an unreadable
  data dump.

Fill in the bracketed dashboards, data sources, and key metrics before deploying. The value of this system comes entirely from data freshness — if any connected source is allowed to serve cached results silently, the core promise (always-current, not a stale export) breaks down.

🔧 Step-by-Step Implementation Guide

1. Data Source Access & Credential Setup Gather API access/credentials for each data source feeding a dashboard — inbox/email account, email marketing platform, competitor tracking source, and CRM/sales system. Confirm each connection can be queried live rather than only via scheduled export, since on-open freshness is the core requirement.

2. Per-Dashboard Data Pipeline Build Build the data pull and normalization logic for each dashboard separately, since each source has a different native format and rate-limit behavior. Keep pipelines modular so one source's API change doesn't break the other three dashboards.

3. Live Artifact Rendering Configure the Claude Artifact rendering layer to trigger a fresh pull on each open rather than displaying a cached render, and design the layout around the key metrics defined for each dashboard so the most important numbers are immediately visible without scrolling or digging.

4. Failure Handling & Monitoring Add explicit handling for data source failures — a dashboard section should show "unavailable" rather than blank or stale data when a connected API doesn't respond. Periodically verify each connection still has valid access, since expired credentials are the most common cause of a dashboard quietly going stale.

⚡ Want This Deployed in 3-5 Days? (Turnkey Option) Don't want to build live API integrations, data normalization pipelines, and always-current dashboard rendering yourself? Get the fully built, tested, and custom-branded Claude Dashboards (Live Artifacts) deployed for your business. 👉 Get Turnkey Deployment at Pole Position Pro Shop

🔗 Related Turnkey Deployments