What's New in Grooper 2026

2026 BETA

This article covers new or changed functionality in the current or upcoming beta version of Grooper. Features are subject to change before version 2026's GA release. Configuration and functionality may differ from later beta builds and the final 2026 release.

This is a placeholder for the Grooper 2026 release. Consider this a work in progress.

Grooper MCP

See full article linked below.

Grooper MCP

Grooper 2026 introduces native support for the Model Context Protocol (MCP). This is the bridge between Grooper and the world of AI agents.

What is an "agent"? An agent is an AI system that can look at a situation, decide what to do next and take action, rather than just answering a single question. Agents call tools to get real work done. Furthermore, it pursues a goal across many steps, deciding its own next moves, checking results and adjusting course as needed.

What are "tools"? A tool is one specific action an agent is allowed to call, such as searching a database, reading a document, or running a calculation. MCP is the common language that lets an agent discover and call tools, including those inside Grooper.

What Grooper tools are accessible using MCP? There are 100+ tools in the Grooper tool catalog. These tools allow agents to create Batches, configure nodes, test extraction, review documents, query statistics and more, all through natural language.

How can MCP benefit me?

New Node Types

MCP Server: With the new MCP Server node, Grooper can now publish its own tools, resources, and prompts to any MCP-compatible client — Claude Code, OpenCode Copilot, or your own agent apps.

MCP Registry: The MCP Registry works in the other direction, aggregating connections to external MCP servers and remote registries so agents and the Agent Planner can reuse third-party capabilities across your repository. Together, these features let you both expose Grooper as a tool for AI assistants and extend Grooper's AI workflows with tools from the broader MCP ecosystem.

Upgrade Notes for 2026

Review these behavior changes we observed in testing before upgrading to version 2026. They may affect existing Projects, so we recommend testing your own documents after the upgrade.

New .NET Libraries

Grooper 2026 is built on a newer version of .NET. This required us to replace many of the libraries the platform uses. In most cases, this improves security and functionality. However, a different library can occasionally produce different results. The differences we observed in testing are listed below.

Tesseract OCR

If you use the Tesseract OCR Engine, you may see slight differences in OCR results. Results are generally improved overall. However, in some of our regression tests, these differences occasionally reduced field extraction when a regex or lexicon-based Extractor was used. Because the overall improvement outweighed these minor differences, we chose not to make additional changes.

If you rely on regex or lexicon-based Extractors with Tesseract, retest them after upgrading.

Auto Orient

The Auto Orient IP Command can use either Tesseract or Transym as the OCR Engine that evaluates page orientation. With the new Tesseract library, we saw differences in how Auto Orient rotated pages.

Tesseract was often less confident in its page orientation assessment when using an orientation-specific model. To address this, we added a fallback mechanism. If the first pass produces a result below Auto Orient's confidence threshold, Tesseract runs a second pass with a different model. If the two passes agree on the orientation, the page is rotated. If they disagree, the page is not rotated.

This approach has trade-offs:

• Pro: Auto Orient should be more accurate than it was before the change.
• Con: The second pass takes additional time, which can slow down an IP Profile that uses Auto Orient. The more pages that require the second pass, the longer Auto Orient takes to run.

Alternatives

If you want to avoid the second pass, you have two options:

• Lower the "Minimum Confidence" value when Tesseract is selected. The second pass only runs when the first pass falls below the minimum confidence, so a lower value means it runs less often. However, a lower value can also cause more false positives. Test any change thoroughly.
• Use Transym as the OCR Engine. We have not observed any differences in Transym's orientation results.