No-code, 3 steps

Name → Starting point → Open. No SQL, no formulas, no "custom views".

26 widget types

KPIs, every chart type, plus advanced viz: heatmap, treemap, waterfall, calendar.

AI everywhere

Create dashboards from a sentence. Chat with your data. Per-tile auto-summary.

Power BI parity

Cross-filter, drill-through, Edit Interactions, conditional formatting, bookmarks.

Setup

Install · configure AI · open the app

Build

Create dashboard · add widgets · arrange

Interact

Cross-filter · Edit Interactions · alerts

Reference

26 widgets · templates · use cases

OpenAI

api.openai.com/v1

Azure OpenAI

Your Azure deployment URL

OpenRouter

100+ models · pay-per-use

Self-hosted

Ollama · LM Studio · vLLM

Colour-coded headers

Sales = green, HR = purple, finance = teal. Match the dashboard to its team at a glance.

Quick-action gear

Rename, share with groups, change refresh interval, delete — without opening the dashboard.

Star & favourite

Pin daily favourites to the top of the kanban so you never hunt for them.

Live refresh badge

"Live 120 s" tells you how often each dashboard refreshes. Configurable per dashboard.

Pick a business area (Sales · Purchase · CRM · Inventory · Accounting · HR · Custom). Only categories whose apps are installed appear.

Pick the Odoo model to read. The recommended choice is starred. Each shows its technical name (sale.order) for clarity.

The wizard groups available fields by kind: numeric (measure), date (filter), group-by. Just for context — no input needed yet.

Type what you want to see in plain English ("Monthly revenue by salesperson") or pick a preset. AI uses this to choose the widget type.

AI proposes the best-fit widget (KPI, bar, treemap, heatmap, …). Override with one click if you'd prefer something else.

Live preview of the actual tile on your real data. Tweak the name, save — and it lands on your dashboard.

Drills straight to the Odoo records list for that segment — useful for "show me the actual orders behind this bar".

Adds a second filter. Combines with the global Period filter and any custom filters you've set.

📈 Value greater than

Useful for "leads in pipeline above 100", "overtime hours above 200".

📉 Value less than

Useful for "revenue below $200k", "stock level below safety threshold".

⚡ % change

Catches sudden swings — "30% revenue drop in one period".

🔄 Per-dashboard interval

Set auto_refresh_seconds on each dashboard. Range 30 s – 1 hour. Quietly cancels itself when the tab is hidden.

⚡ Smart cache invalidation

The two-layer cache keys on the underlying model's max write_date. Add a sale order and the affected widgets refresh on next poll — no stale data.

⏸ Pause on inactivity

When the browser tab is hidden, polling pauses automatically. Resumes on focus. Saves server CPU on tabs no one is watching.

KPI Tile

Big number. One aggregate value with optional delta vs. previous period and conditional formatting.

Best for: headline metrics, scorecards, summaries.

Configure: model + measure (sum/avg/count) + optional target + compare-period.

Bullet / Progress

Value vs. target. Horizontal bar that fills as you approach the goal.

Best for: sales targets, campaign goals, OKRs.

Configure: set kpi_target.

Progress Ring

Donut showing % of target. Same as bullet but circular.

Best for: compact target tracking on KPI strip.

Gauge

Speedometer. Min/max bracket with a needle indicator.

Best for: capacity-bound metrics (server load, fulfilment).

Configure: gauge_min and gauge_max.

Bar chart

Compare categories. Vertical bars; great for ranking.

Best for: revenue by rep, orders by status, items by category.

Stacked bar

Two dimensions at once. X-axis primary, each bar split by secondary group.

Best for: revenue by month, split by product category.

Horizontal bar

Long category names. Same as bar but rotated — labels readable.

Best for: customer rankings, country lists.

Line chart

Trend over time. Smooth line through data points.

Best for: monthly orders, weekly leads, daily traffic.

Stepped line

Discrete trend. Right-angle steps emphasise "constant between samples".

Best for: headcount, subscribers, contract count.

Area chart

Filled trend. Visual emphasis on magnitude. Includes optional forecast.

Best for: cumulative revenue, stock value over time.

Pie chart

Share of total. Each slice is one category's contribution.

Best for: order status split, payment-method breakdown.

Doughnut chart

Pie with a hole. Same purpose; centre lets you show the total.

Best for: companies vs. individuals, channel split.

Polar area chart

Radial slices. Each segment has the same angle; the radius is the value.

Best for: visual comparison across 4-10 categories.

Radar chart

Multi-attribute "shape". Each spoke is a category; polygon shows the profile.

Best for: employee skills, product mix, customer segments.

Scatter chart

2D correlation. Each record at its (x, y) values to reveal clusters and outliers.

Best for: revenue vs. discount, lead score vs. conversion.

Bubble chart

Scatter with a third dimension. Bubble size encodes another value.

Best for: deal value (size) by lead score (x) and age (y).

Mixed (bar + line)

Two metrics, one chart. Bars + a line on twin Y-axes.

Best for: revenue (bars) + order count (line).

Funnel

Drop-off across stages. Each bar narrows as records progress.

Best for: lead → opportunity → quotation → sale.

Pivot / Table

Numbers, not pictures. Group by rows × split by columns.

Best for: finance reports, audit trails.

Top-N list

Leaderboard with bars. Top N by your chosen measure; each row has an inline progress bar.

Best for: top customers, top products.

Records list

Real Odoo records. Click a row to open the record in Odoo.

Best for: recent orders, leads needing follow-up.

Rich-text panel

Narration tile. Embed instructions, notes, links.

Best for: "About this dashboard" intros, contact info.

🔥 Heatmap (matrix)

2D grid where colour intensity = value. Spot patterns instantly.

Configure: set both primary AND secondary group_by fields.

Example: orders by status × month.

🌲 Treemap

Rectangles sized by value. The biggest contributors take the biggest area.

Configure: standard group_by + measure. Optional limit.

Example: revenue by customer.

📅 Calendar heatmap

One year, GitHub-style. 53 weeks × 7 weekdays, each cell colour-coded by daily activity.

Configure: set group_by_field to a date or datetime column.

Example: daily order activity, ticket volume.

💧 Waterfall

Sequential contributions to a total. Each bar starts where the previous one ended.

Configure: typical sum + group_by_date.

Example: monthly revenue building to YTD.

Create with AI

Describe what you want to track. The AI reads your installed Odoo apps + data model and proposes a complete dashboard.

Trigger: top menu → ✨ Create with AI

Chat with the dashboard

Open the AI panel and ask any question about the current view. The AI reads live tile data and answers.

Trigger: dashboard toolbar → AI chat

Per-tile AI insight

Right-click any tile → ✨ AI insight. Get a 1-paragraph summary with notable outliers called out.

Trigger: tile kebab (⋮) → ✨ AI insight

High-level visibility, zero noise

Headline numbers for the board meeting: revenue vs. target, AR aging, top customers, monthly trend. One screen, no scrolling.

Pipeline you can actually act on

Sales-rep leaderboard, win-rate funnel, top customers, pipeline by stage. Cross-filter to any rep and the whole dashboard pivots.

Headcount, leaves, attendance

Employees by department, leaves by month, attendance heatmap, top departments. Installs in one click.

P&L, cash flow, AR aging

Revenue waterfall by month, expense breakdown, top vendors, cash flow trend. Compare-period KPIs catch dips fast.

Campaign ROI & funnel performance

Lead → opportunity → won funnel, campaign spend vs. budget, top-converting sources, monthly trend.

Inventory, purchasing & manufacturing

Stock value, low-stock alerts, top movements, PO status, manufacturing throughput, helpdesk team load.

⚡ Query layer

• Native read_group for every aggregate — pushed down to PostgreSQL as a single GROUP BY
• No record loading — Python never instantiates browse records
• Indexes respected — uses your existing PG indexes on date and FK columns
• Single round-trip per widget — no N+1 queries
• Concurrent widget loads — the dashboard fetches multiple widgets in parallel

💾 Cache architecture

• Layer 1: per-widget LRU — in-process, ~2,000 entries, TTL configurable per widget
• Layer 2: data-version key — auto-invalidated when underlying records change
• Sharded by user — each user's row-level security applied before caching
• Domain-aware — cross-filters and Period filter are part of the cache key
• Manual override — per-tile "Refresh" bypasses both layers

🏢 Multi-company

• Per-company AI configuration — each company can use its own API key
• Standard company_id domain on every model — chart data respects the active company filter
• Dashboards can be assigned to specific companies or made shared
• Cross-company aggregations work when the user has access to multiple companies

🌐 Internationalisation

• Every UI string translatable via standard Odoo i18n (.po files)
• Number formatting via Intl.NumberFormat — respects user locale
• Date axis labels in the user's language
• Currency formatting on KPI tiles via user's preferred currency
• RTL languages (Arabic, Hebrew) — layout flips correctly

♿ Accessibility

• All buttons keyboard-focusable with visible focus rings
• Semantic HTML — headings, lists, tables for screen readers
• Chart tooltips also surfaced as text on hover / focus
• Sufficient colour contrast (WCAG AA) in both light and dark themes
• Tile menus operable without a pointer device

🌙 Light / dark theme

• Automatically follows Odoo's user theme (light / dark / auto)
• All 26 widgets re-skin cleanly in dark mode
• Chart palettes adjusted for both themes
• Per-dashboard custom accent colour overrides theme accents

🎨 5 chart palettes

Pick a palette per widget:

• Vivid — default, bright colours
• Ocean — cool blues / teals
• Sunset — warm oranges / yellows
• Forest — greens / earth tones
• Mono — black / grey / white

🌈 Per-dashboard accent

Each dashboard has its own accent colour applied to:

• Kanban card header strip
• KPI tile primary number colour
• Progress bars and rings
• Header left border
• Cross-filter pill accent

🖌 Per-tile icon + colour

Each tile lets you pick:

• FontAwesome icon (any in the catalogue)
• Header icon colour
• KPI value colour
• Override the dashboard's palette per tile

🧩 Add a custom widget type

1. Inherit smart.dashboard.item in your module
2. Add your widget type to WIDGET_TYPES
3. Register the OWL component in WIDGET_COMPONENTS
4. Provide a _compute_<type> method on the data engine

🪝 Hooks & extension points

• _build_domain — add custom filters server-side
• _serialize_for_client — add fields to the JSON payload
• paletteColors() — define your own palette
• smart_dashboard.template — register your own templates
• JSON-RPC endpoints — fetch widget data from external apps

⏱ 0:00 — Open Smart Dashboards → ✨ New Dashboard

Name it "Monday Sales Review". Pick 📦 Use a template. The catalogue opens.

⏱ 0:30 — Pick "Sales Performance+" template

10 widgets land: Total revenue KPI, Pipeline funnel, Revenue by salesperson bar chart, Top 10 customers list, Monthly revenue trend, Quote conversion rate, Recent orders list, plus a few more.

⏱ 1:30 — Add 2 custom widgets

Click + Add widget. Through the 6-step wizard:

• Heatmap of sales by salesperson × month (3 minutes including layout adjustments)
• Waterfall showing month-by-month revenue contributions building to YTD (2 minutes)

⏱ 4:30 — Configure conditional formatting on Total revenue

Edit the Total revenue tile → set alert high = 110% of target, warn low = 80% of target. The tile now paints itself green/amber/red automatically each Monday.

⏱ 5:00 — Create an alert

🔔 Alerts → New → "Weekly revenue under 80% of target" → email me + the sales lead. The cron checks every 30 min. We get a Slack-style ping the moment things drift.

⏱ 6:00 — Done. Bookmark the Monday view.

Set Period = "Last 7 days", click Views → + Save current view → name it "Monday standup". Every Monday at 9 AM, one click loads the exact state needed for the meeting.