Pathway, a female-founded AI startup, has launched its data processing engine, which is up to 90x faster than existing streaming solutions. The engine combines batch and streaming logic in the same workflow, allowing the system to learn and react to changes in real-time. This approach has traditionally divided batch and streaming architectures, slowing down AI system adoption and affecting their intelligence. Pathway CEO and co-founder Zuzanna Stamirowska emphasizes the added complexity of generative AI workflows.
Stamirowska highlights the need for rapid data processing and adaptable AI, enabling real-time processing for developers using batch, streaming, or LLM systems. AI retraining is crucial for machines to forget outdated information in real-time, as traditional methods require retraining.
ChatGPT questions whether AI language models can unlearn information if they are inaccurate. OpenAI developers can update and retrain the model based on new data and improvements. The pathway can make revisions to data points without requiring a full batch data upload, similar to updating the value of one cell within an Excel document. Pathway’s clients, DB Schenker and La Poste, have reduced anomaly-detection analytics projects’ time-to-market from three months to one hour, and fleet CAPEX reductions of 16%.
Pathway, founded by Polish-French duo Stamirowska and Claire Nouet, is a female-led deep tech startup raising $4.5mn in a pre-seed round. The female-led startup aims to become a “lingua franca” for data pipelines, cutting costs for clients and democratizing developers’ design of streaming workflows.
