\n\n
All-in-one agentic diagramming

Draw your diagramswith AI agents.

Describe it below and get an editable diagram in seconds — or connect the agent in your editor and it draws straight from your repo. Data platforms, cloud infrastructure, system design.

0/2000• 3 credits left
Try:

No sign-up required • Free to try

Works with your agentImageClaude CodeImageCursorImageCopilotImageCodexImageGemini CLI
AI-native, not AI-flavoured

Your agent can draw the diagram

A real MCP server. Claude Code, Cursor or Codex creates, edits and exports diagrams, reading your repo for context.

one-time setup
claude mcp add datadef https://datadef.io/mcp --header "Authorization: Bearer dd_live_YOUR_KEY"

then just ask

“Draw our Snowflake platform with Fivetran and dbt”
“Add a Kafka topic between the API and the bronze table”
“Read this repo and diagram how the services talk”
“Export that diagram and drop it in the README”
claude — datadef mcp
Get your API key

Free to start · Works with any MCP client

Or automate it.
Your repo, documented and diagrammed, daily.

Connect the repo, pick a branch or a tag. Datadef draws the diagram, writes the doc, and keeps both current.

1 · Connect

2 · Generate

3 · Embed

ImageGitHubImageGitLabImageAzure DevOpsread-only · branch or tag · daily sync
Repo to diagram

From prompt to diagram in seconds

Plain English in. An editable canvas out: nodes, zones, labelled edges, lineage.

1

Describe your stack

Plain English. No syntax.

2

AI generates everything

Nodes, zones, icons, edges.

3

Edit with full control

Drag, resize, refine with AI.

Share & deliver

PNG, link, or live embed.

Under 2 minutes to first diagram
Generate your first diagram

Free to start · No credit card

Speak the same language

From code nobody reads to diagrams everyone gets

Drag the handle. The raw file on one side, what your team actually reads on the other.

Revenue

The metric everyone quotes, documented from models/marts/fct_revenue.sql.

Lineage

The same metric shown as a Datadef diagram with column-level lineage

Definition

select
  d.month_key,
  c.segment,
  sum(o.net_amount) as revenue
from {{ ref('stg_orders') }} o
join {{ ref('dim_customer') }} c on c.customer_sk = o.customer_sk
join {{ ref('dim_date') }} d on d.date_key = o.order_date
where o.order_status = 'fulfilled'
group by 1, 2

Notes

  • Net revenue, recognised monthly. Only fulfilled orders count.
  • The lineage above is embedded from the canvas, so it stays current.
# Revenue
 
The metric everyone quotes, documented from `models/marts/fct_revenue.sql`.
 
## Lineage
 
![Revenue lineage](https://datadef.io/api/embed/revenue-lineage)
 
## Definition
 
```sql
select
d.month_key,
c.segment,
sum(o.net_amount) as revenue
from {{ ref('stg_orders') }} o
join {{ ref('dim_customer') }} c on c.customer_sk = o.customer_sk
join {{ ref('dim_date') }} d on d.date_key = o.order_date
where o.order_status = 'fulfilled'
group by 1, 2
```
 
## Notes
 
- Net revenue, recognised monthly. Only fulfilled orders count.
- The lineage above is embedded from the canvas, so it stays current.
 
docs/metrics/revenue.mdrendered
Raw Data
Transformations
KPI Definitions
Dashboards

Complete KPI Journey

See exactly where each metric comes from, how it's calculated, and where it's used.

Column-level

lineage, not just table-to-table

Business + Technical

One view for PMs (“what it means”) and engineers (“how it’s built”).

Trust

between teams

Impact Analysis

Know what breaks downstream before you change anything upstream.

Fewer surprises

bad decisions avoided

Map source to warehouse to KPI

Trace any metric to its raw source in seconds.

Column-level lineageLink repos and dashboardsOwners and teams

1-3h saved weekly · 10x faster deliverables

Send a link, skip the meeting

A 30-second share replaces the 30-minute walkthrough.

No viewer accountThey get it instantlyRevoke anytime, the link dies
Real-time collaboration with multiple team members editing the same canvas

Work together, live

Share a link, everyone edits live. No setup, no sync hassle.

Draw.io / Lucidchart

Drive sync, version conflicts

Datadef

One link, live cursors

Up to 10 collaborators per projectCursors and changes in real time
100% free · no account

Try the full editor,
right now, free.

The scratch canvas is the real product. Draw, generate, edit. It saves in your browser.

Drag and connectFreehand notesTables and KPIs500+ tech icons
Open Scratch Pad

No account • No credit card • Saves locally

The Datadef editor with a node selected and its properties panel open

Join 500+ data professionals

From solo consultants to enterprise teams — real architectures built with Datadef

Modern ELT stack — Fivetran, Snowflake, dbt and Looker
Databricks medallion lakehouse on Azure
Retail star schema — Kimball dimensional model
Real-time streaming analytics — Kafka, Flink and ClickHouse
Modern ELT stack — Fivetran, Snowflake, dbt and Looker
Databricks medallion lakehouse on Azure
Retail star schema — Kimball dimensional model
Real-time streaming analytics — Kafka, Flink and ClickHouse
Customer 360 — identity resolution and activation
ML feature platform — offline training and online serving
Data quality and observability layer
IoT telemetry platform on AWS
Customer 360 — identity resolution and activation
ML feature platform — offline training and online serving
Data quality and observability layer
IoT telemetry platform on AWS

Simple pricing, no surprises

Start free with Scratch Canvas. Upgrade when you need cloud storage, exports, and team features.

Most popular

Pro

For individual data professionals

$12/month

Billed annually ($144/year)

Free for 7 days at signup, no card needed

Team

For data teams & organizations

$99/month

Up to 5 seats · flat price · billed monthly

Create the workspace and invite people free — $99/month when you activate

Every new account starts with 7 days of everything, no card. Cancel anytime, no questions asked.

Frequently Asked Questions

What types of data diagrams can I create?

Datadef specializes in data architecture: pipelines, warehouses, lakehouses, ETL/ELT flows, data lineage maps, KPI dashboards, and system architectures. Every diagram uses data-specific components like tables, pipelines, data sources, and KPI indicators.

How does the AI generation work?

Describe your data stack in plain English: “E-commerce data platform with Fivetran, Snowflake, dbt, and Looker.” Our AI generates a complete, editable diagram with properly connected components, zones and column-level lineage — usually in under two minutes.

Can I edit AI-generated diagrams?

Yes! All AI-generated diagrams are fully editable. Drag nodes, add components, modify connections, change colors, add metadata. The AI gives you a head start — you have full control to refine everything.

How does lineage tracking work?

Datadef automatically generates lineage from your diagram connections. Click any KPI or table to see its upstream sources and downstream consumers. Define formulas, track transformations, and do impact analysis before making changes.

What export formats are supported?

Export as PNG, JPEG, or PDF for presentations. Export JSON for version control and data portability. Share public links with stakeholders who don't need to log in.

Is my data secure?

Datadef never connects to your actual databases. You describe your architecture — we never access your data sources or store sensitive information. All data is encrypted in transit and at rest.

Stop wasting time on diagrams

Whether you're a freelancer delivering client work or a team lead managing complexity — let AI generate the diagrams while you focus on what matters. Every new account starts with a 7-day free trial of everything, no card needed.

No credit card requiredUnder 2 min to first diagram