Version 2.1
Web app
- Added Phone Upload with one-time QR pairing and encrypted direct or relay transfer to the desktop. BiomAPI handles connection setup only and never receives the selected file bytes.
- Unified buttons, workflow panels, file previews, modal actions, and results surfaces into a more consistent responsive visual system.
- Reworked the biometry input and results displays for improved usability and more efficient use of space.
- Added an optional first-time walkthrough covering uploads, result verification, BiomPIN privacy and storage, History, and BiomAPI integrations.
- Redesigned the main biometry input around Process, BiomPIN, and History, keeping single-file upload as the default while moving multiple-file upload and manual transcription into the file selector.
- Added a redesigned multiple-file workflow with Slow and Standard processing, usage estimates, quota checks, BYOK guidance, optional spreadsheet creation, optional local history, and clearer progress and results.
- Simplified results metadata so provenance and processing details are easier to scan.
- Added a responsive comparison view for reopening the original uploaded PDF or image alongside extracted or edited results.
- Added thumbs-up/down extraction feedback, with optional follow-up messages.
- Added display-only derived keratometry, including delta K and delta PK magnitudes and axes.
- Added a BiomPIN identity selector for excluding patient initials and ID from copied links and QR codes.
- Added this in-app changelog.
API
- Added a Browser SDK with Web Widget integration for BiomAPI’s full upload and review interface, PIN and link sharing, URL-fragment handling, initials/ID context, and optional local history. The utilities also work with custom REST integrations.
- Added exam-date extraction across the API, web interface, editor, CSV output, CLI reference, and documentation.
- Simplified public extraction metadata while retaining model, BYOK, timestamp, and provenance information.
- Added request IDs to successful response metadata.
- Added an extraction-feedback endpoint accepting only a rating and the original processing request ID.
- Removed internal token, retry, timing, and preprocessing diagnostics from public API payloads.
- Added the optional Slow service tier for PDF and image processing at 0.5 BiomAI credit, with exact half-credit quota accounting and Standard remaining the default.
- Replaced the standalone CSV endpoint with a configurable ZIP export that can include matching CSV and XLSX spreadsheets and individual JSON files.
BiomAI
- Upgraded BiomAI to Gemini 3.7 Flash and aligned its extraction flow with current Gemini API best practices.
- Added deterministic preprocessing to improve performance and extraction accuracy.
- Made the Gemini connectivity health check lighter and more reliable.
- Added per-user BiomAI credential policies for disabling server-managed extraction, using the deployment default, or routing to assigned AI Studio or Vertex AI credentials.
- Added cached engine reuse and separated assigned-credential usage from shared deployment capacity.
- Added global and public-global BiomAI quotas alongside per-user and per-IP rate limits.
CLI
- Expanded CLI results with BiomAPI retrieval links and ready-to-use ESCRS IOL Calculator links.
- Released CLI and Agent Skill 2.1 with safer local credential storage, collision-safe downloads, ordered four-worker batch processing, a 15 MiB upload limit, structured errors, and improved Codex and Claude Code discovery and installation.
- Added Slow service-tier selection and combined CSV, XLSX, and JSON export, and aligned client behavior with deployment-configured upload limits and processing timeouts.