Kili Technology
kili-technology.com
- Free trial
- Web
- API
- Self-hosted
What Kili Technology is
Kili Technology is an expert-in-the-loop platform for creating, reviewing, and evaluating AI data, including training and ground-truth datasets. It handles images, video, PDF documents, text, audio, and geospatial data. Teams can configure multi-step, consensus, sequential, and cross-review workflows, then inspect agreement and accuracy by class or person, honeypot calibration, project analytics, and an Explore view. Annotation options include bounding boxes, polygons, semantic segmentation, pose estimation, named entity recognition, and classification. Kili connects with AWS, Azure, Google Cloud Platform, and S3-compatible storage. Automation is available through its API, Python SDK, plugins, and MCP server. Deployment choices include SaaS, a managed Azure Marketplace application, and on-premise installation. Kili states that it has SOC 2 Type II, ISO 27001:2022, and HIPAA certifications. A two-week trial offers full access, with one seat and specified asset limits. Paid pricing is available on request.
Who it suits
Kili suits organizations coordinating annotation across domain experts, internal teams, and external workforces. It is relevant where review controls, cloud connections, deployment choice, and dataset exports matter.
Good to know, and what to watch
- Supports images, video, text, audio, PDFs, and geospatial data
- Configurable multi-step and consensus review workflows
- API, Python SDK, plugins, and MCP automation
- SaaS, Azure Marketplace, and on-premise deployment
- Includes agreement, accuracy, and project analytics
- No free plan beyond the limited trial
- Grow and Enterprise pricing requires contacting Kili
- Free Trial limits seats and asset counts
- Advanced use cases require contacting the team
Plans
Every plan and priceEnterprise
Grow
Free Trial
Pricing on request · 14-day trial
The verdict
Kili provides a broad annotation and review environment for teams that need structured quality checks and deployment flexibility. Confirm asset volumes, seat needs, and pricing during the trial or sales process.
What the maker publishes
- Access controls
- The platform describes project-level data isolation, role-based access, encryption at rest and in transit, and audit trails. source
- Company history
- Kili Technology says its founders established the company in 2018 and that its platform went live in July 2020. source
- Data types
- The platform supports images, video, PDF documents, text, audio, and geospatial data. source
- Deployment
- Kili is offered as SaaS, as a managed application in Azure Marketplace, and as an on-premise deployment. source
- Founded
- 2018 source
- Integrations
- Kili connects to AWS, Azure, Google Cloud Platform, and S3-compatible storage, and supports automation through its API, Python SDK, plugins, and MCP server. source
- Intended users
- Kili says the platform is designed for organizations managing annotation across domain experts, internal teams, and external workforces. source
- Plan limits
- The Free Trial lists one team seat, up to 100 text, document, and image assets, and up to 5 video and satellite imagery assets; advanced use cases require contacting the team. source
- Purpose
- Kili is an expert-in-the-loop platform for creating, reviewing, and evaluating AI data, including training datasets and ground-truth datasets. source
- Quality tools
- Kili provides per-class and per-person agreement and accuracy metrics, honeypot calibration, project analytics, and an Explore view. source
- Review workflows
- Teams can configure multi-step review, consensus labeling, sequential review, and cross-review workflows. source
- Security
- Kili states that it has SOC 2 Type II, ISO 27001:2022, and HIPAA certifications and undergoes annual audits. source
- Support
- The Grow plan says its support level is adapted to customer needs, and lists professional services as available. source
- Trial evaluation
- Kili offers teams a free two-week evaluation with full platform access to test on their own data. source
Checked against the maker's pages on 7 Oct 2026.