Agent Name: Geotab Ace
Goal: Deliver customized fleet data insights and provide answers to general support questions using natural language
Version: 2.1.0
Developed By: Geotab AI Engineering team
Date of Release: July 14, 2026
How can I use this product? Ace is embedded in MyGeotab. Also, as an end user you just have to:
Geotab Ace is a Generative AI (GenAI) assistant that enables users to interact with complex fleet datasets through natural conversation, democratizing data access by removing technical barriers.
Target users: Fleet managers with access to MyGeotab.
Out-of-scope uses: This system will only answer questions and give information related to fleet management or general support questions.
Supported languages: Approximately those available through the Gemini web app. Ace uses a suite of Gemini models (the main model is Gemini 3.5 Flash), for which no standalone public language list is published.
Purpose-built for fleet management: Integrated with the MyGeotab platform, Ace allows fleet managers to access performance data and actionable insights through simple conversations.
Automated SQL generation and data retrieval: Ace translates natural language questions into precise SQL queries, executes them, and delivers customized information that meets each user's specific needs.
Context-aware retrieval: The agent prioritizes retrieving relevant, up-to-date information directly from the user's fleet database rather than relying solely on pre-trained knowledge.
Integrated Support Assistance: Beyond data retrieval, Ace provides answers to general support questions, helping users navigate features and troubleshoot technical or operational issues.
Chart Generation: When query results are suitable for visualization, Ace automatically generates charts within the chat response, helping users interpret fleet data at a glance without any additional steps.
Map Generation: For location-based data, Ace renders interactive map views directly in the chat response, allowing fleet managers to visualize geographic patterns and activity spatially.
Ace is built on a LangGraph multi-agent system. A Brain agent handles orchestration and routing across three specialized sub-agents, each running its own processing loop.
Brain Orchestration: The Brain agent interprets the confirmed question and intent, then routes to the appropriate sub-agent. Sub-agents are called sequentially, one at a time, with each result building on the last.
Query Sub-agent: Handles data questions by generating a SQL query, executing it against the user's fleet database via BigQuery, and iteratively refining the query if errors occur. Each delegation runs its own refinement loop.
Support Sub-agent: Handles general support questions by searching Geotab's knowledge base to surface relevant help content.
Response: Ace returns the answer along with its reasoning and, where applicable, the underlying query and any charts or map visualizations, so the user can verify the result.
Logging: Interactions and operational metadata are logged for monitoring, quality control, and audit purposes.
Brain agent: Coordinates intent understanding, routing, reasoning, and retries across the system.
Language models: A fleet of Gemini models, each assigned to a specific role, including orchestration, SQL generation, reasoning summarization, insight extraction, lightweight routing tasks, and schema and question retrieval. All model calls are routed through Geotab's internal ecosystem.
Data platform & storage: Geotab's Data as a Service (DaaS) platform provides the scalable infrastructure and distributed backbone for high-performance data access, while storage systems manage fleet data, conversation context, and system logs.
Fleet users are restricted to accessing data solely from their own organizations.
Note: If you aren't sure about what is included, ask Geotab Ace "What data do you have access to?" and check back regularly, as new data is continuously being added. In addition to understanding what data is available, this question will also provide data history. While data history may vary by table, all data is available from January 1st, 2024 and onwards.
Knowledge Constraints: Ace is intentionally constrained to content relevant to user's questions. This constraint follows the data minimization principle specified in the General Data Protection Regulation (GDPR).
The agent generates its chain-of-thought, reasoning process, and SQL query to the user.
This documentation and Geotab's Responsible AI site are efforts towards Geotab's commitment to transparency in its responsible AI practices.
Red team testing was conducted prior to release to proactively identify vulnerabilities, adversarial inputs, and potential misuse patterns; findings informed the guardrails and output filters applied to the system.
Guardrails have been added to restrict irrelevant topics to fleet management and enforce compliance.
The RAG-based architecture grounds the responses as much as possible to prevent hallucination.
Responses are filtered to reduce harmful outputs.
Audit logs track and review generated responses for quality control.
Ace, as an agent, is responsible for generating and executing queries. Results are stored in a bucket with access granted solely to the user interface and the API for display to the customer. Customer data never leaves Geotab's GCP infrastructure during Ace query processing.
Ace utilizes external closed generative AI models designed with strict data privacy controls. While relevant customer data is processed by these models to provide better services, it is never used to train or refine the underlying AI models. Contractual terms with Google via Google Enterprise Agent Platform explicitly prohibit the use of customer data or user prompts for model training. These closed models maintain a separation between the data provided and the models' training data, ensuring that the data remains private. Moreover, the Ace development team is advised to follow the data minimization practices, ensuring that only the minimum information necessary to fulfill a request is shared with these external models.
This metric indicates how closely the LLM-generated SQL query aligns with the expected query.
Evaluates the performance of our SQL generation by comparing the results of the generated queries against the expected query results. The score considers both the accuracy of the data retrieved (matching rows and columns) and the completeness of the results (missed or additional rows).
Shows how accurately the LLM-generated SQL queries match the correct database tables. It directly measures the SQL's success in targeting the correct tables based on the input query.
Tracks the number of SQL queries the LLM generated before producing a final result. A higher number of attempts could indicate difficulty in understanding or constructing the correct query.
Feedback Mechanism: Users can provide feedback on the Ace interface using a thumbs-up or thumbs-down rating system. The AI Engineering group performs a weekly manual analysis of the feedback.