Turn messy data into clean models
Meet Chonk, your AI guide to dbt. Whether you're new to dbt or scaling an existing project, Chonk helps you build production-ready models, tests, and docs — the right way.
// THE CHALLENGE
dbt is powerful. Learning it is hard.
You know dbt is the answer. You just need help getting there.
Raw data everywhere
Inconsistent column names, missing documentation, and no clear structure.
No time to learn
dbt has a steep learning curve. You need results now, not weeks from now.
Unsure where to start
Staging model or dimension? What tests? The decisions are overwhelming.
Technical debt
Quick fixes become permanent. Every change risks breaking something.
From chaos to clarity
FROM prod.raw_tbl_usr_2024
WHERE usr_stat = 'A'
-- TODO: what does A mean?
-- NOTE: ask Dave about col_23
AND dt_crt > '2024-01-01'
-- FIXME: hardcoded date
# TODO: add documentation
# NOTE: what columns are important?
models:
[]
Which model depends on which? Who knows.
SELECT * FROM {{ source('raw', 'users') }}
),
renamed AS (
SELECT
id AS user_id,
usr_stat AS status,
dt_crt AS created_at
FROM source
)
- name: stg_users
description: "Cleaned user data"
columns:
- name: user_id
tests: [unique, not_null]
- name: status
tests:
- accepted_values:
values: ['active', 'inactive']
Clear lineage. Every model documented and tested.
Smart enough to know what he doesn't know
Chonk learns from every project. He already knows thousands of common patterns, so he only asks when there's real ambiguity. And once you tell him, he remembers forever.
Chonk
Your data-sniffing corgi
*sniffs database* Found 47 tables. Already recognized 38 common patterns from my training. Just need your help with a few things...
I see user_status, status, and user_state - are these the same thing, or does user_state mean something different here?
user_state is for US states (billing address). The other two are the same.
Got it! I'll remember that for your whole org. Already building your staging layer with 12 models, 47 tests, and full docs...
Asks Only What Matters
Chonk already knows common patterns. He only asks when there's genuine ambiguity that needs your business context.
Learns & Remembers
Answer once, and Chonk remembers - across your project, your org, and even globally for common patterns.
You Stay in Control
Chonk suggests, you decide. Every choice is yours to confirm, modify, or override.
Three levels of memory. Chonk learns patterns globally (common column names), at your org level (your company's conventions), and per project (specific decisions). The more you use him, the smarter he gets.
// WHO IT'S FOR
Built for people who need results, not complexity
Whether you're just starting out or scaling your data team, DataChonk meets you where you are.
Startups & Small Teams
No dedicated data engineer? No problem.
You're a founder, analyst, or developer who needs to set up proper data infrastructure without hiring a $200k/year analytics engineer.
- Get production-ready dbt projects in minutes
- Follow best practices from day one
- Scale confidently as your data grows
Learning dbt
The best way to learn is by example.
You're studying dbt but struggling to connect the concepts. DataChonk generates real, working code you can learn from and modify.
- See best practices in action
- Understand why models are structured this way
- Build your portfolio with real projects
Data Teams
Accelerate your analytics engineering.
You know dbt but spend too much time on boilerplate. DataChonk handles the scaffolding so you can focus on business logic.
- Generate staging layers in seconds
- Consistent patterns across your project
- Auto-generate documentation as you build
Consultants & Agencies
Deliver more value, faster.
You're building data infrastructure for clients. DataChonk helps you deliver polished, documented dbt projects in a fraction of the time.
- Impress clients with comprehensive docs
- Standardize your delivery across projects
- Focus on strategy, not scaffolding
Everything you need to build great dbt projects
Stop wrestling with boilerplate. DataChonk combines AI expertise with dbt best practices to accelerate your analytics engineering workflow.
Chonk-First Architecture
Modular building blocks for your data warehouse. Source, Staging, Entity, Fact, Metric, Docs — each piece is reviewable and self-documenting.
Expert dbt Brain
An AI that actually understands dbt — from basic refs to advanced incremental strategies, semantic layer, and warehouse-specific optimizations.
Production-Ready Output
Generate complete dbt projects with proper structure, naming conventions, tests, and documentation that looks like a senior AE wrote it.
Package Intelligence
Smart recommendations for dbt_utils, dbt_expectations, elementary, and more. Know exactly why and where each macro is used.
Living Documentation
Auto-generated docs blocks, column descriptions, exposures, and governance notes. Your documentation stays in sync with your code.
Ship to dbt
Push to GitHub, GitLab, Bitbucket, or Azure DevOps — then trigger dbt Cloud runs automatically. Build in Chonk, ship to production in one seamless flow.
Always Up to Date
Knowledge sync keeps DataChonk current with the latest dbt releases, best practices, and documentation changes.
Get better at dbt, faster
DataChonk helps you learn and use dbt the right way — whether you're preparing for your first dbt init or scaling an established data platform. Build with confidence, ship to dbt Cloud or Core.
New to dbt
Learn dbt by building real projects
You know you need dbt but don't know where to start. DataChonk teaches you dbt best practices while helping you build — so you learn the right patterns from day one.
The Scenario
Your startup just got product-market fit and data is piling up. You need a proper analytics stack yesterday, but you've never touched dbt before.
The DataChonk Way
- 1Connect to your warehouse (Snowflake, BigQuery, Redshift, etc.)
- 2Chonk explains why each model layer exists as you build
- 3Generate proper staging, marts, tests, and docs — with explanations
- 4Ship to dbt Cloud or Core — ready for production
Without DataChonk
Reading docs, watching tutorials, and hoping you're doing it right. Weeks of learning before you ship anything useful.
Why DataChonk?
DataChonk helps you learn and use dbt the right way — from your first model to production scale.
DataChonk enhances your dbt workflow — we help you build better dbt projects, not replace dbt. Works seamlessly with dbt Cloud and dbt Core.
// THE CHONK SYSTEM
Modular building blocks for your data warehouse
Each chonk is a distinct, reviewable unit of analytics engineering work. Think of them as Lego bricks for your dbt project.
Source Chonk
Define your raw data sources with freshness checks
What's Included
- Source definitions with database/schema
- Freshness configuration
- Column-level documentation
- Primary key tests
Output Path
sources/<source_name>.yml1version: 223sources:4 - name: stripe5 database: raw6 schema: stripe_prod7 freshness:8 warn_after: {count: 12, period: hour}9 tables:10 - name: customers11 columns:12 - name: customer_id13 tests: [unique, not_null]// EXPERT-LEVEL OUTPUT
Code that looks like it was written by a senior AE
DataChonk generates production-ready dbt code with proper incremental strategies, surrogate keys, and warehouse-specific optimizations. Every model includes comprehensive tests and documentation.
Supported Warehouses
1{{2 config(3 materialized='incremental',4 unique_key='order_key',5 incremental_strategy='merge',6 on_schema_change='append_new_columns'7 )8}}910with source as (11 select * from {{ ref('stg_shopify__orders') }}12 {% if is_incremental() %}13 where updated_at > (14 select coalesce(max(_loaded_at), '1900-01-01')15 from {{ this }}16 )17 {% endif %}18),1920transformed as (21 select22 {{ dbt_utils.generate_surrogate_key(['order_id']) }} as order_key,23 order_id,24 customer_id,25 total_amount_cents / 100.0 as total_amount,26 order_status,27 updated_at as _loaded_at28 from source29)3031select * from transformedLet Chonk sniff out your data mess
Chonk transforms raw tables into clean, documented, tested dbt projects. You focus on insights — he handles the modeling.
