Scrunch ships real-time AI search features faster with Tinybird

Scrunch is a Series A venture-backed startup at the forefront of a new category: AI search optimization. As AI-powered search engines like ChatGPT, Perplexity, and Google AI Mode reshape how people discover information, Scrunch helps marketing teams understand and optimize how their brands, products, and services appear in these new platforms.

Building a data-intensive platform for the AI search era

For Robert McCloy, CTO and co-founder of Scrunch, building a data-intensive application was a given from day one. The core of Scrunch's value proposition requires ingesting massive amounts of data about how companies show up in AI search engines, analyzing crawl patterns, and delivering actionable insights to help customers grow their presence in AI search.

Like many startups, Scrunch initially built their application using standard tools: a transactional database for app data, and GCP Pub/Sub and BigQuery for streaming data processing and analytics. While these technologies are powerful and appropriate for many use cases, they proved to be heavyweight for the kind of real-time features Scrunch needed to build.

A new feature needed real-time visibility... fast

The catalyst for Scrunch's journey with Tinybird came while building Agent Traffic, a feature that provides real-time visibility into crawler activity on customer websites. AI search engines like ChatGPT use various user agents to crawl websites, and these crawls have a direct impact on how companies appear in AI search results.

The data loads for Agent Traffic are highly variable. Some sites receive minimal crawler traffic, while others are constantly being indexed. More importantly, Scrunch's customers wanted to see this data in real time - to watch as ChatGPT or Perplexity crawled their site live, making the connection between AI search activity and their content visceral and immediate.

The team at Scrunch knew their current data infrastructure wouldn't deliver that real-time experience. They considered building classical stream processing pipelines, but that felt like an enormous engineering lift for a startup focused on rapid iteration and market validation.

From evaluation to prod in 48 hours

As Scrunch started looking for a solution to their real-time data challenges, Tinybird stood out for one critical reason: developer experience.

What impressed the Scrunch team was how end-to-end Tinybird's solution was. Instead of stitching together multiple systems - event ingestion, stream processing, data storage, and API layer - Tinybird provided all of it in a single, cohesive platform. Raw website access logs could be ingested natively, transformed through SQL-based pipelines, and exposed as API endpoints ready to power their application.

The proof of concept moved remarkably fast. From the decision to try Tinybird to having a working prototype in production took just 48 hours. Getting the MVP of Agent Traffic up and running took about a day. For a startup where velocity is everything, this was transformative.

Solving the classic BI trade-off

Across much of Scrunch's application, they face a challenge common to many analytics platforms: the trade-off between freshness and performance. Pre-summarized data is fast but not real-time. Raw data is fresh but slow to query at scale.

Traditionally, solving this requires significant engineering effort - building materialized views, definining indexes, managing refresh schedules, and creating complex systems to fuse aggregate and real-time data. For each new feature or filter, the complexity multiplies.

Tinybird's approach of defining pipelines and endpoints solves this fundamental problem. The cost of adding another aggregation or roll-up is minimal, and there's much less maintenance overhead for the development team. Features that would have taken weeks to build with traditional infrastructure can be implemented in days.

Cost predictability at startup scale

For Robert and the Scrunch team, cost predictability matters as much as performance. Having worked extensively with analytics platforms that charge per byte read or per compute, he's experienced the pain of surprise bills - those distressing conversations with the CFO about why a missing WHERE clause resulted in a $20,000 data warehouse bill.

With Tinybird, costs have been linear with usage. There's no worry about accidentally triggering an expensive full table scan or forgetting to optimize a query. For a startup that's tripled in size since adopting Tinybird, this predictability is invaluable.

The future: Saying yes to more features

As Scrunch looks ahead to the next quarter and beyond, the ability to move quickly on new data features will be a competitive advantage. In the fast-moving AI search optimization market, no one knows exactly what the future holds. Being able to try things out, make multiple bets, and iterate quickly is crucial.

There are features that Scrunch simply wouldn't build without Tinybird. If each feature takes multiple sprints, you have to pick and choose. But if those same features take days instead, suddenly the calculus changes. The team can say yes to more customer requests, explore more use cases, pull in those QoL updates, and deliver more value.

For sophisticated customers in a rapidly evolving market, this velocity matters. While product discipline and saying no are important, the ability to say yes - to actually deliver the features customers need - can make all the difference in winning a new category.

What's next for Scrunch and Tinybird

Scrunch currently uses Tinybird primarily for the Agent Traffic feature, but the team sees opportunities to adopt it more broadly across their application. Many of the data-intensive features on their roadmap - better summarization, insights, different roll-ups, and more sophisticated filters - are exactly the kind of capabilities where Tinybird excels.

As Scrunch continues to grow and the AI search optimization market matures, Scrunch sees the opportunity to grab - and secure - market share by shipping what their customers really need faster than their competitors. With Tinybird, they have a platform that lets them move fast, control costs, and focus on building customer value rather than managing data infrastructure.