play
Watch demo(1 min)
SQL cell
comment
app-builder
auto-run
Atriedes DB
chevron-down
Browse schema
GENERATE
close
select order_date,
sum(number_of_items) as total_items
from prod.dim_orders
group by 1
order by 1 desc
output-arrow
dataframe
Introducing

Hex Magic

AI-powered tools for humans doing amazing things with data.

Public Beta
Get started

Do what you do best, let the computer do the rest

Magic features are fully-integrated into the Hex workspace, putting powerful AI assist in every SQL and Python cell.

SQL

Query with a click

Hex understands your schemas and project context, so it can help with everything from quick questions, to auto-completing joins, to generating a finicky date filter.

Complex patterns
Error debugging
Finicky syntax
Query from scratch
SQL cell
comment
app-builder
auto-run
Atriedes DB
chevron-down
Browse schema
GENERATE
close
select order_date,
sum(number_of_items) as total_items
from prod.dim_orders
group by 1
order by 1 desc
output-arrow
dataframe
Python + R

Scripting superpowers

Obscure syntax got you down? Hex knows your packages inside and out — all you have to do is ask.

Pandas made accessible
Magic error fixing
Deep knowledge of packages and your data
Python cell
comment
app-builder
auto-run
GENERATE
close
data
SQL cell
comment
app-builder
auto-run
Atriedes DB
chevron-down
Browse schema
GENERATE
close
select cast(date_trun('month', orders.ordered_at) as date) as month,
category,
is_spicy,
sum(order_details.price) AS order_total,
count(distinct orders.order_id) as count
from prod.dim_orders orders
left join prod.order_details order_details on order_details.order_id = orders.order_id
output-arrow
dataframe
bug
Debug

Tricky bug? Hidden tpyo? Hex can help.

Missing parentheses, botched imports, and plain bad code don’t stand a chance. One click, magic fix.

magic
Magic fix
arrow-left
Click that button
Python cell
comment
app-builder
auto-run
GENERATE
close
model = ensembleModels.RandomForestClassifier(
random_state = 222
)
 
upsample = SMOTE(random_state = 111)
scaler = StandardScaler()
 
features_names = [col for col in data.columns if col != 'Churn']
features = data[feature_names]
scaled_features = scaler.fit_transform(features)
target = data['Churn'].to_numpy()
r_features, r_target = upsampler.fit_resample(scaled_features, target)
code-comment
Explain

No more mystery code

Quickly explain and document code, whether it's something you wrote, or... borrowed... from the internet.

"Magic has been super helpful here because I'm not a Python expert, so I typically rely on Magic to help refactor/edit code that I'm not familiar with."

Olivier Dupuis

Discursus

"I would have had to go and spend 2 hours digging through Stack Overflow to find where to start. Magic spun up something very close, very quickly. And then we were able to iterate on it super fast. It's something that we wouldn't have been able to without Magic."

David Selwood

Block Renovation

"It's a time saver not having to write perfect syntax all the time, 'how do I get all the unique rows versus if it's a single column' is hard to remember."

Akshay Bhat

Notion

Powerful AI for creative humans

At Hex, we believe in human-computer symbiosis. We should partner, with each of us focusing on the things our respective neural nets are uniquely good at.

Hex Magic is built to augment and accelerate human insight. AI models don't know the right questions to ask, or how to present the answers – that’s your job! And we think it should stay that way.

Now Available in Public Beta
Get started

FAQ

Is my data being sent to a third party?
chevron-down

We’re not sending contents of dataframes anywhere, so no – your actual data never leaves Hex. We do, however, use metadata about the project context, including upstream code, database schemas, and your prompt text. If any of these contain sensitive information, or you’re not comfortable with that, you should not use Hex’s Magic features.

What’s the quality of the completions?
chevron-down

Pretty good! But it does make mistakes, both obvious and subtle. We make no guarantees that your code will be correct, or insights accurate. You are responsible for the code you execute. Please read more on our Learn page.

How do I get access?
chevron-down

We're currently in Public Beta, so if you're excited to try experimental features you can turn it on via your Settings page.

Are you going to charge for this?
chevron-down

Probably! Right now we’re focused on making Hex feel amazing and [whispers so investors can’t hear] not worrying too much about the money. Magic gets expensive, though, so expect there to be some reasonable charge down the road.

Can't find your answer here? Get in touch.