Reaches #1 News App ranking in major markets with Google Cloud
Cuts AI processing costs by multiples to fuel growth with Gemini in Vertex AI
Unlocks access to niche language news sources with Speech-to-Text
Eliminates infrastructure overhead to accelerate innovation with Vertex AI
Centralizes user behavior insights to improve retention with BigQuery
Syft used scalable AI to bring global news to free thinkers, slashing costs and propelling its app to #1 in the process.
Syft used scalable AI to bring global news to free thinkers, slashing costs and propelling its app to #1 in the process.
The cost of escaping the matrix
Syft exists to help users "leave the Matrix"—local echo chambers and filtered news—and access unbiased, hyper-personalized insights from around the world. To deliver on this promise, the app must ingest, translate, and analyze thousands of diverse sources daily, from major international outlets to niche blogs.
However, early iterations of the platform relied on third-party large language models (LLMs) that proved prohibitively expensive. The cost of running deep research on every news channel created a financial barrier that would have forced the company to restrict access to a wealthy niche, directly contradicting its mission to democratize information.
Attempting to solve this, the engineering team briefly experimented with building their own infrastructure. They procured GPUs and deployed open-source models, hoping to bring operations in-house. But this shift came with the complexity of managing the "plumbing." The team found themselves spending valuable cycles debugging hardware and maintaining uptime instead of improving the product.
They realized they needed an infrastructure that functioned like a public utility—always on and instantly available. They sought an on-demand AI solution: a resource they could simply tap into to power their innovation, rather than a complex generator they had to build, maintain, and fuel themselves.
If we continued with our previous providers, we would have had to charge hundreds of dollars per user, which is the opposite of democratizing information. We asked ourselves: Why reinvent the wheel when Google Cloud already provides the efficiency and scale we need?
Robert Zheng
Co-founder, Syft
Decoding world news with semantic AI
When a user defines their interests in Syft—whether it's "European macroeconomics" or "US Politics"—the app initiates an elaborate, real-time hunt for relevant insights.
Moving beyond simple keyword matching, Syft uses Gemini in Vertex AI to ingest thousands of global sources and decipher the semantic intent of every article. For example, if a major financial story breaks in Paris, the model identifies the story's relevance to a user’s specific economic interests, even if the article uses local colloquialisms or indirect phrasing that legacy keyword search engines would miss.
We use Gemini in Vertex AI because no other model can semantically understand what a user truly wants in the same way. It allows us to go beyond keywords and deliver the fundamental meaning of a story, regardless of the language it was written in.
Robert Zheng
Co-founder, Syft
The architecture handles diverse formats and languages through a unified Google Cloud ecosystem:
Audio Transcription: For news broadcasts in niche languages such as Malayalam—where other providers struggled with accuracy—Syft utilizes Speech-to-Text to accurately transcribe and index audio content, ensuring no perspective is lost due to a language barrier.
Data Centralization: To track engagement, user behavior data from Google Analytics and Firebase flows directly into BigQuery. This acts as the central hub for decision-making, allowing the team to spot organic "snowball effects" in new markets and adjust their strategy immediately.
Under orchestration by Vertex AI, Syft's AI newsroom can deliver hyper-personalized daily digests at a fraction of the cost of its previous infrastructure.
Slashing costs to fuel the climb to #1
By moving to Gemini in Vertex AI, Syft reduced its AI processing costs by multiples compared to previous providers. The capital saved was reinvested directly into user acquisition campaigns, propelling Syft to become the #1 News App on Google Play in major countries, surpassing established social media giants.
With Google Cloud handling the infrastructure, the engineering team stopped managing servers and started shipping features. This freedom allowed them to launch a new AI-narrated podcast mode in just weeks, designed for the large cohort of users who listen to news on the go.
Looking ahead, the team plans to deepen its relationship with Google Cloud by exploring scalable vector search solutions to further refine its retrieval capabilities. They are also experimenting with next-generation Gemini models to generate dynamic web widgets, ensuring that as the platform grows, their infrastructure remains as agile as their news feed.
There wouldn't be Syft without the Gemini LLMs. We simply wouldn't be able to offer something this impactful in a cost-effective way without the Google Cloud ecosystem supporting us.
Robert Zheng
Co-founder, Syft
Syft is an AI-native news company that aggregates and analyzes thousands of global sources to deliver "pure signal" news feeds. By using advanced semantic analysis, it eliminates clickbait and language barriers, providing users with a hyper-personalized, multi-perspective view of world events.
Industries: Media and Entertainment, Startup
Location: Singapore
Products: Gemini in Vertex AI, BigQuery, Speech-to-Text, Firebase, Google Analytics