Thaw converts photographs of a board into editable pen strokes within a notebook, combining handwriting tools with AI interpretation. Its official description also connects lecture audio, page checks and an Obsidian route. This first look uses the App Store listing, not a classroom trial or a test of mathematical accuracy.
The audience is a student or note-taker who wants photographed material to become something they can revise rather than a fixed image. The important boundary is between the source, the AI's reading and the edited notes. An editable equation can be useful while still being wrong, and a confident correction should remain open to comparison with the original material.
Based on official store information. We have not tested this app hands-on.
What it does
A board photo or up to thirty library images can be processed into headings, text, equations, tables and diagrams. Blocks appear beside the relevant strip of the original board, with difficult readings marked for review. Inserted material uses a chosen pen and color. That source alignment is particularly useful as a review concept, though we have not tested its reliability.
The developer describes a second reading of equation crops and code-based checking of board mathematics, with apparent board errors noted in red rather than silently replaced. Page checks also provide reasons beside proposed corrections. These are claims about an AI-assisted workflow, not our verification that every equation, diagram or inferred correction is right.
Writing tools include pressure- and tilt-responsive pens, highlighting, erasers, a lasso and shapes. Inserted ink can be edited as text or through a maths keyboard, and pencil corrections can be interpreted. PDFs can be annotated, images pasted and pages exported. Notebook writing is designed for Apple Pencil on iPad, with finger drawing available through Settings.
Before you install
Begin with a small board whose content you understand. Compare the original strip with each interpreted block before relying on the inserted notes. We have not tested recognition, stroke editability, notation coverage or mathematical reasoning. A red correction is an AI proposal with a stated reason, not an authority that overrides the teacher or the source.
Lecture recordings can be captured on iPhone or added from Files. The listing says transcription happens on the device, then Claude uses the material to supplement notes with marked provenance. Recording permission and an institution's rules still matter. Obtain the appropriate consent rather than treating an app feature as permission to record another person's teaching or voice.
The data boundary is explicit: board interpretation, mathematical checks and several other AI operations use Anthropic's Claude after consent. On-device audio transcription is a different stage and does not make the whole AI workflow local. Review what the actual request includes before processing sensitive material. The developer warns that Claude can be wrong.
Price and access
The retained App Store offer lists a free download. AI features use credits, with a board described as about one credit and a class's notes about five. A new device receives free monthly credits. These are approximate usage descriptions, not a promise that every input has an identical cost.
Thaw Pro increases the allowance and adds Claude Opus through monthly or annual subscription access. Exact regional prices and allowances are not established in the retained description. Confirm the current purchase terms and device requirements. Ordinary writing, AI access and suitable hardware are separate parts of the arrangement, so a free installer should not be treated as unlimited AI processing.
Our take
Thaw's strongest proposition is a link between captured evidence and material that can actually be edited. A board strip beside the result can support checking, while provenance for lecture additions can help distinguish what came from where. Those boundaries matter more than an unqualified promise that the resulting mathematics can be trusted.
We have not tested accuracy, accessibility, privacy, sync or recovery. Our take is to validate one known page with its original retained, then inspect the editable result and export. Choose the AI features when their errors can be identified and corrected in your workflow, not simply because a photo becomes polished ink.
The Obsidian route saves a class as Markdown and brings edits back, according to the listing. That can support another working environment, but we have not tested conflict handling or preservation of every diagram. Keep a representative sample before treating two-way editing as lossless in every case.
Cross-device use also has a specific shape. The iPhone captures and supports reading or search, the iPad is the main handwriting surface, and the Mac can read, edit and answer questions about notes. Private iCloud sync or direct Wi-Fi between owned devices are described, not a guarantee of instant availability in every network arrangement.
The qualified first-look verdict is an ambitious notebook that should be judged by traceable corrections and usable source context. It can support a thoughtful learning process only when the human reader remains able to question the transformation, preserve the original and recognize that an AI explanation may still need outside verification.
Sources
App icon from the official store listing.
