Annota AI

Annota AI is a workspace for image annotation, document OCR review, and AI-assisted data preparation. It fits solo practitioners and small teams handling both visual datasets and document text; projects limited to one of those jobs may be better served by a narrower tool. The free tier is useful for a small start, while collaboration and higher-volume work require paid capacity.

Overview

Annota AI combines image and document workflows in one product. Image projects support PNG and JPEG files, with bounding boxes, polygons, and segmentation masks. Document datasets accept PDF, PNG, and JPEG files, with page-by-page OCR extraction for review and correction.

That combination is the product's main practical distinction. A team preparing training images and cleaning scanned documents can keep both tasks in one workspace instead of maintaining separate systems. Conversely, an image-only labeling project may not benefit from the document tooling.

Key features

Magic Select generates an object mask that users can inspect, while Magic Annotate finds selected labels across an image or dataset. These model-assisted operations can reduce repetitive labeling, but they do not remove the need for human checking. Manual tools let users draw and edit boxes and polygons, and exports include images and annotations in COCO format, a useful handoff for workflows built around that dataset structure.

OCR processes document content one page at a time, allowing corrections before export. A Markdown ZIP suits general text pipelines; an Obsidian-compatible LLM Wiki workspace is aimed at users who organize knowledge in Obsidian-style structures.

Compatible AI agents can connect through MCP using remote servers with Streamable HTTP and OAuth. Agents inherit existing organization permissions, and billable operations consume the same credits as activity in the app. That keeps access governance consistent, but automated usage can draw down a shared allowance quickly.

YOLO object-detection training remains in private beta and needs separate organization approval; subscribing to a paid plan does not provide access. Brain, the proposed private knowledge base for questions across document datasets, is coming soon rather than part of the current offering.

The maker says core infrastructure and AI models run in EU regions and that AI features do not use external AI-provider APIs. Workspace permissions control access, connected app authorizations can be revoked, and SOC 2, ISO 27001, and ISO 42001 programs are in progress.

Pricing

Annota AI has a free plan and paid plans from €18/mo in its pricing note, with the displayed Pro rate at €21.96 a month including 22% VAT. There is no free trial.

• Free — free: one member, two datasets, 4 GB storage, and €4 in monthly credits. It is suitable for an individual evaluating the workflow or running a small project, but the dataset, storage, and credit ceilings make sustained or varied work restrictive.
• Pro — €21.96 a month: one member, unlimited datasets, 20 GB storage, and €30 in monthly credits; billed with 22% VAT included. This is the logical solo upgrade when dataset count is the main constraint, though it remains a single-seat plan.
• Team — €29.28 a month per member: minimum two members, 40 GB storage per member, €30 in monthly credits per member, shared datasets, and member roles; 22% VAT is included. It adds genuine collaboration, but the minimum commitment means it costs more than Pro even for a two-person operation.
• Enterprise — custom pricing: a quote for the installation, with private-cloud, on-premise, and air-gapped deployment options. This targets organizations with deployment or isolation requirements rather than ordinary project teams.

Platforms

Annota AI is offered as self-hosted software and a web application. Enterprise deployment options extend to private cloud, on-premise, and air-gapped environments.

Who it's for

Choose it when one workflow needs both assisted image labeling and OCR-based document review, particularly as a solo user on Pro or a small group that needs shared datasets and roles on Team. EU-region infrastructure and organization-level permissions may matter to teams with explicit access-control or data-location requirements.

It is less compelling for image-only work, for users needing generally available YOLO training, or for anyone expecting Brain now. The free tier is best treated as an evaluation allowance rather than a long-term team workspace.

Pros and cons

Pros

• Two data-preparation modes in one workspace: image annotation and page-level document OCR cover different but commonly connected tasks.
• Useful export routes: COCO for image datasets, plus Markdown ZIP and an Obsidian-compatible workspace for documents.
• Governed agent access: MCP connections use existing permissions, with revocable authorizations.
• Flexible deployment at Enterprise: private-cloud, on-premise, and air-gapped options address stricter environments.

Cons

• Small free allowance: two datasets, 4 GB, one member, and €4 in credits limit ongoing use.
• Collaboration raises the bill: Team requires at least two members and charges per member.
• YOLO training is not assured: private-beta access requires separate approval, regardless of plan.
• Brain is not ready: the planned document knowledge base cannot currently be relied on.

Alternatives

Label Studio is the better choice for a broad, established platform mix including API, desktop, self-hosted, and web use, with both free and trial access.

MakeSense.ai suits readers wanting a free, browser-based photo-labeling tool with no paid tier.

Roboflow is worth considering when a free tier with 10 monthly credits and a stated model-training and inference allowance matches the project.

VGG Image Annotator is a free option with no paid tiers or usage limits stated.

Vailabel Studio is free and open source, with no account required.

CVAT offers a free MIT-licensed community plan and a hosted free tier for small-scale use.

Draw Attention is aimed at interactive images, with a free edition and a paid single-site license.

Labellerr provides a free researcher plan with 2,500 data credits and one seat, while its Pro plan is priced for larger commitments.

For a wider directory of options, see Image Annotation Software.

Verdict

Annota AI is a strong fit for individuals and small teams that genuinely need image annotation and document OCR review together, especially when COCO and document-oriented exports matter. Its main reasons to look elsewhere are the limited free capacity, per-member Team pricing, and the fact that YOLO training and Brain are not broadly available.