Frequently Asked Questions¶
What is Falkonry Time Series Intelligence (TSI)?¶
Falkonry Time Series Intelligence (TSI) is an AI-native engineering platform designed to automatically detect anomalies and patterns in metrics, telemetry, SCADA, and IoT data. Powered by Falkonry’s globally patented PatternIQ™ technology, TSI specializes in time series visualization, anomaly detection, classification, and collaborative analysis.
Built for digital teams, TSI eliminates the need for extensive IT expertise, enabling users to act on intelligent alerts from thresholds, anomaly scores, and early warning patterns. Its purpose is to accelerate the adoption of data-driven operations by providing a secure, reliable, and cost-effective decision support system.
With TSI, organizations can rapidly convert raw time series data into actionable insights - without manual coding or infrastructure overhead - empowering teams to improve reliability, efficiency, and operational outcomes.
Click here to learn more about the motivation behind TSI.
Click here to learn more about the overall architecture of TSI.
Click here to learn more about the different deployment options supported by TSI.
How does Falkonry address the challenges of analyzing observability data?¶
Analyzing observability data is difficult due to irregular timestamps, gaps, noise, outliers, large volumes of operational data, data flow reliability issues, constant evolution in operations, lack of known examples, and cost/performance constraints. Falkonry addresses these challenges through its AI-native TSI tools by:
- Organizing large volumes of time series data for ease of use.
- Allowing visual review of numeric and categorical data and their trends.
- Finding metrics, time ranges, and patterns that affect systems.
- Performing live data analysis with deployable models.
- Offering a domain-agnostic, open architecture with flexible deployment options (Falkonry-hosted cloud, private cloud, or on-premise "air gap" mode).
- Providing robust integration support through REST APIs and ready-to-use data connectors for various sources like AWS IoT Core, Azure IoT, iba Systems, Litmus, and MQTT.
I'm new to Falkonry TSI and have received an invite to join Falkonry TSI. I'm unable to log in. How do I proceed?¶
For the first time acces, users must set a strong password using the "Sign Up" options on the login screen.
What is the Falkonry Bootcamp, and what certifications are offered?¶
The Falkonry Bootcamp aims to upskill industry professionals with AI for advanced continuous improvement. It offers a two-level certification process:
- Falkonry Specialist: A half-day, self-paced, virtual program covering Falkonry TSI Fundamentals, Data Management, and Automation.
- Falkonry Professional: A half-day, self-paced, virtual program focusing on AI topics like detecting anomalies, exploiting patterns, rules, guards & alerts, analytics strategy, and a capstone project.
How does TSI handle data ingestion and connectivity?¶
TSI provides flexible and secure inbound connectivity options to bring in both live and historical time-series data. It integrates with various industrial data sources like PLCs, SCADA, historians, and IoT devices using industry-standard protocols such as MQTT, Parquet, and CSV. Falkonry operates on a "customer-push" model, meaning it never initiates connections to a customer's OT/IT systems; instead, customers push their data to Falkonry.
What data formats does TSI support?¶
TSI primarily supports two signal types: Numeric and Categorical.
- Timestamp - Unix Epoch and ISO 8601. The "time" column should consistently be named "time" and timestamps should be normalized to UTC
- Numeric data - US locale number format
- File structure - Files should be strictly rectangular, with consistent column names (case-sensitive)
- File formats - Parquet is the most efficient and therefore recommended. CSV is suitable for small test data files (under 10 MB)
How are signals managed and organized in Falkonry?¶
In TSI, when data is ingested, the system automatically detects whether a signal is Numeric or Categorical and places it in a draft state. Data for signals in draft state is not processed until they are explicitly approved by a user. Users have the opportunity to review and change a signal's data type while it is in draft mode. Once a signal is approved, its data type cannot be changed.
Signals can be organized into hierarchical trees.
What are Falkonry Rules and how are they used?¶
A Rule in TSI is a set of criteria applied to one or more signals that results in a True or False assessment. These rules are evaluated automatically and periodically by Falkonry, and they can be configured to generate alerts for tracking and notification of important events.
How does Falkonry support integration with other systems (outbound connectivity)?¶
TSI provides flexible outbound connectivity options to send live outputs to external systems, enabling push notifications and integration with various maintenance management solutions. System generated output can also be accessed/exported using supported REST API.
How are users and access managed in Falkonry?¶
Access to TSI is strictly by invitation only. See User Management for details.
What are best practices for modeling in Falkonry Patterns?¶
Effective modeling in Falkonry Patterns involves several best practices across data preparation, model creation, and refinement.
What precautions should be taken while sending historical data over a connection?¶
Efficient data loading and management are crucial for optimal performance in Falkonry TSI. Historical data should not be uploaded into a live connections or sent over MQTT connections.
What are the different types of AI models in Falkonry Patterns, and how do they learn?¶
Falkonry Patterns utilizes various AI models to discover patterns from time series data, namely, Unsupervised Learning, Semi-Supervised Learning, or Supervised Learning. These models, typically of the Sliding Window type, continuously analyze data within defined observation windows to produce assessments and identify conditions.