Discovery engine
for your 

company

app

website

company

Recommendation APIs that increase your metrics 🚀

hero-iphone-discoveryhero-desktop-discovery

Trusted by investors from

Leverage Shaped across multiple industries and use-cases

With Shaped, you can build powerful ranking models that understand the content your users are most likely to engage with or buy. You can use this understanding to create personalized discovery pages, feeds, recommendations and marketing communication that your users will love.

Marketplaces
usecase-marketplaces-recommendations-grocery
Grocery Delivery
usecase-marketplaces-recommendations-sneakers
Fashion
usecase-marketplaces-recommendations-furniture
and more...
For you page
usecase-social-recommendations-newsfeeds
Feeds
usecase-social-recommendations-socialmedia
and more...
usecase-ecommerce-recommendations-recommendedforyou
Recommended for you
usecase-ecommerce-recommendations-similartoyou
Similar to you
usecase-ecommerce-recommendations-jacketsforyou
and more...
usecase-media-video-streaming
Video
usecase-media-recommendations-news
News
usecase-media-recommendations-podcasts
and more...
usecase-other-recommendations-recruitmentleads
Recruitment leads
usecase-other-recommendations-productivitytools
Productivity tools
usecase-other-recommendations-settings
and more...

Designed for developers

We wrapped all the ranking infrastructure you need into an elegant, easy-to-use API. That means you don’t need to spend months handling all the data engineering, model training, and deployment difficulties inherent to recommendation systems.

# The Shaped model endpoint
shaped=https://api.shaped.ai/v0/models
# Request results for your ranking model
curl "$shaped/rec/rank?user_id=1" \
--header 'x-api-key: DEADBEEF'
# Your response is returned as a json list
{
"rankings": [
"item413",
"item255",
"item338",
]
}
developers-iphone-realestate
Houses you might like

Leaders use ranking to build world-class discovery experiences

The last decade has seen a shift from search to discovery-led products. It has become an expectation that products learn and adapt to user behavior. Ranking is the technology that has driven this shift.

  • leaders-youtube-recommendations

    70%

    Watch time from what the algorithm recommends

  • leaders-tiktok-for you

    52mins

    Average per day per user
    on ‘For you’ page

  • leaders-amazon-recommendations

    30%

    Percentage of revenue
    from recommendations

  • leaders-airbnb-recommendations

    40%

    Relative increase in
    bookings revenue

  • leaders-netflix-recommendations

    75%

    Content discovered
    from recommendations

  • leaders-instagram-recommendations

    90M

    Personalization predictions
    per second for the explore tab

In the news

No items found.

Testimonials

Coming from building ad personalization at Facebook and then leading ecommerce data science at Thrasio, I know how hard it is to bake personalization into products. Shaped is the first approach that truly builds personalization as a product, served through an API that easily plugs in to your team’s existing products. There is simply no better way to optimize your search, recommendations, or marketplaces than leveraging industry-leading data science practices using Shaped.

Maddison Hawkinson
Ex-Head of Data Science and Machine Learning at Thrasio

Been using Shaped for the last 6 months and have gotten a 20% increase in engagement. We consider them a valuable part of our infrastructure – we’d otherwise have to build out a bespoke ML stack which we aren’t equipped to do as a small team. Got my first model in a day and it took less than 2 weeks to deploy to prod. Team is really helpful and quick to respond to any issues. Highly recommend them. Didn’t want to put my name to this to prevent our competitors from seeing.

Anon
Founder and CTO

With the TikTok-ification of everything, personalization has become incredibly important for all companies to get right. I love what Shaped is doing to democratize the insanely complex ranking systems that are used at companies like Facebook and Youtube. It really does level the playing field so us founders, product managers and engineers can focus on our core differentiation.

John Barnett
Product Lead at Twitter. Previously CEO at Chroma Labs (acquired by Twitter), Product at Instagram & Facebook.

We considered hiring an ML team to build ranking but it was costly and scoped to take 10-12 months. After a vendor search we decided to go with Shaped because we didn’t have to mess around setting up new data pipelines to send our data to 3rd parties and they’re not a black box  — our engineers have control over the set-up of the models and the results.

Mishaal Al Gergawi
Co-founder and CEO, Axis

How Shaped works

1

Connect your data stack with our API

2

Shaped trains and deploys your models

shaped-train-deploy-machine-learning
3

Call our API to get a list of ranked results

Rank results
4

Increase your revenue

Increase business metrics

Modern machine-learning ranking stack

Created by the AI researchers and engineers that built the ranking models for Facebook. Our mission at Shaped is to make state-of-the-art-deep learning infrastructure accessible to more companies.

less data required

Less data required

Shaped pre-trains models so you don’t need much data to get started. This is what allows you to get great results from day 1.

less data required

Use all data types

Shaped drives growth with user interaction data and unstructured data types like images, videos, text or social graphs.

less data required

Deploy in hours

Shaped is built directly on top of your data stack meaning we can train and deploy predictive ranking models in hours.

TwitterLinkedIn

Join the Shaped community

Listen to what experts are saying
Daanish Khazi
@bertgodel

@shapedai has got to be the most compelling product I've seen in years - will prob bootstrap the next dozen billion-dollar consumer apps. 🚀

Kudos to the team @tulliemurrell @shaibruhis @heydaniel_c

Jacopo Tagliabue
@jacopotagliabue

Modern Data Stack and RecSys for the Reasonable Scale: what's not to like?

Shaped brings a fresh developer-first perspective to an evergreen ML problem.

Sammy Sidhu
@Sammy_sihu

I love that @shapedai has opened up each of the 4 stages of the recommender system via API. Other products are a black box. No idea what’s going on inside.

Shaped is designed exactly how developer first ML products should be built 🙌

Mike Williams
@Yoroomie

I just met up with @heydaniel_c and got to learn more about what they’re building with @shapedai, which helps marketplaces with search & ranking.

It’s great to see such high caliber teams focused on building tools for marketplaces.

Sahil Madan
@hillatio

Just tried out @shapedai and it’s awesome 🔥

Easy to get a model set-up without pushing data anywhere and great docs

John Barnett
@johnbarnettt

Shaped is democratizing these powerful tools so that any company can now leverage cutting edge AI to serve their customers.

ML in production is now 10x faster and easier.

Mishaal Al Gergawi
@algergawi

@shapedai has reduced the time to add ranking to my product by 99%.

Made it super easy for me to experiment with things I would have never even tried to do before 🤯

Rachael Wood
@racheldowns26

We rolled our own recommendations system with my last marketplace, but using a product like Shaped is like having a whole ML team in your corner instead. Huge superboost for early platforms.

Farbod Motlagh
@FarbodTM

@shapedai is a super flexible API that gives developers full control over ML recommendation systems 👌 Love it!

Ready to get started?

For developers

Waitlist for public API keys 🔑

You're on the waitlist!
We'll be in touch 🙌

Oops! Something went wrong while submitting the form.

For companies

Let's schedule a demo with your data ⚡

Thanks for signing up!
We'll be in touch 🙌

Oops! Something went wrong while submitting the form.

FAQs

How long will it take to integrate?

You can get a model created in a few hours. Integrating that model in typically takes 2-4 days.

How is Shaped better than Algolia, AWS Personalize...?

For the aforementioned companies, recommender and personalization solutions are an afterthought to their core business. Whereas for Shaped, it is our specialty. We use state-of-the-art machine-learning techniques and pre-train our models so customers don’t need as much data to get started.

Can I deploy multiple machine-learning models with Shaped?

Absolutely. In fact we encourage it. Modern tech companies have implemented 10s-100s of unique ranking models in their products for various use-cases. Each model usually requires a team of people to build and maintain. You can create an additional model with just one request.

Why shouldn’t I build this in-house?

The high cost of hiring machine-learning engineers, the long time required to build and the on-going full-time maintenance. Shaped can take you from 0 to 1 in a few days at a fraction of the cost.

What’s your pricing?

Our pricing is a usage based flat-fee per month. Once we know your approximate number of users and items we can provide an estimate.

How do you handle my data?

Shaped doesn’t store your data. We fetch, train and immediately discard it. Great care has been taken to ensure the critical paths are secure. Shaped is SOC 2 Type 1 in progress.

Built by scientists, engineers, designers, and product from