Interfaze
An AI startup focused on deterministic AI tasks. It provides models with verifiable outputs for document processing, OCR, object detection and web scraping, serving primarily the healthcare, legal and financial industries.
Timeline (last 90 days)
Interfaze handles 50 million API requests per month, and its entire backend (including model APIs, code sandboxes, and vendor routing) runs on Vercel, with no separate Kubernetes cluster or dedicated operations team.
Interfaze focuses on deterministic tasks (such as document processing, OCR, object detection, web scraping, speech-to-text, and classification) rather than general-purpose conversational models.
When Interfaze's models return results, they include bounding boxes (indicating where the numbers came from) and confidence scores, making the output verifiable, while still accepting prompt input.
Both Interfaze's frontend and backend run on Vercel, all existing as serverless functions in the same codebase.
When a request requires computation, Interfaze hands the code to Vercel Sandbox for isolated execution.
Vercel's Fluid compute carries Interfaze's Puppeteer-based web scraping workload, which has demanding memory and timeout requirements.
Interfaze uses Vercel AI Gateway to route data cleaning tasks among open-source models hosted by multiple third-party vendors, and AI Gateway automatically fails over when a vendor fails or is rate-limited.
Interfaze uses the Vercel Workflow SDK to build its lifecycle email system, leveraging durable execution to enable workflows to pause for months and resume with full state.
In Interfaze's release process, a new model version corresponds to a new branch, the Vercel-side build takes about three minutes, and then the preview URL is sent to a few customers for a month of testing before merging.
Interfaze recently released Interfaze 1 Lite, an open-weights version that can run on a single GPU and is served through the same API.
1 more timeline events
All Facts (11)
Interfaze handles 50 million API requests per month, and its entire backend (including model APIs, code sandboxes, and vendor routing) runs on Vercel, with no separate Kubernetes cluster or dedicated operations team.
50%UnverifiedInterfaze focuses on deterministic tasks (such as document processing, OCR, object detection, web scraping, speech-to-text, and classification) rather than general-purpose conversational models.
50%UnverifiedWhen Interfaze's models return results, they include bounding boxes (indicating where the numbers came from) and confidence scores, making the output verifiable, while still accepting prompt input.
50%UnverifiedBoth Interfaze's frontend and backend run on Vercel, all existing as serverless functions in the same codebase.
50%UnverifiedWhen a request requires computation, Interfaze hands the code to Vercel Sandbox for isolated execution.
50%UnverifiedVercel's Fluid compute carries Interfaze's Puppeteer-based web scraping workload, which has demanding memory and timeout requirements.
50%UnverifiedInterfaze uses Vercel AI Gateway to route data cleaning tasks among open-source models hosted by multiple third-party vendors, and AI Gateway automatically fails over when a vendor fails or is rate-limited.
50%UnverifiedInterfaze uses the Vercel Workflow SDK to build its lifecycle email system, leveraging durable execution to enable workflows to pause for months and resume with full state.
50%UnverifiedIn Interfaze's release process, a new model version corresponds to a new branch, the Vercel-side build takes about three minutes, and then the preview URL is sent to a few customers for a month of testing before merging.
50%UnverifiedInterfaze recently released Interfaze 1 Lite, an open-weights version that can run on a single GPU and is served through the same API.
50%UnverifiedInterfaze's next plan is to launch a web search gateway, with one API routing to the search vendor chosen by the developer, similar to what AI Gateway does for models.
50%