On October 14, 2026, Google Research announced the public launch of Deep Review for analysing complex documents. For a small engineering or consulting team, the launch offers a reason to test whether an additional AI critique can expose inconsistencies before a document reaches a client.
What is available
Google says users receive two free document reviews each week. Initial inputs are PDF and plain text; reviewing several files together is still a planned capability. The company describes the tool as an aid for checking reasoning, claims and consistency.
A condition for business documents
Google says uploaded documents are not used to train or improve its models, but are stored for one year for display to the user. These are separate conditions: no training does not mean no retention.
How a small business could evaluate it
Start with your own non-confidential technical document that the team has already checked manually. Compare the AI’s comments against the original: which errors did it find, which did it miss, and which supposed problems did it invent? Separately count the time spent assessing its recommendations.
This is Nadali’s proposed evaluation approach, not the result of a Nadali test. Do not upload client material, personal information or trade secrets without checking permissions and processing terms. Our data-boundaries guide can help frame that decision.
Verification limits
Availability and features are reported from Google’s announcement. Nadali has not tested the service: the product page supplied no readable text to our verification tool. Its findings require human review; an AI assessment neither certifies a document nor replaces professional judgment.
Bottom line: evaluate the new tool as an additional critique, with separate attention to errors, staff time and data retention.
