Artificial Intelligence Technologies and Intelligent Automation Solutions

The Vendor is required to provide to modernize and responsibly expand the use of Artificial Intelligence (AI) across internal operations and public-facing services.
- The County is exploring AI capabilities across multiple domains, including: form automation and completeness checks; plan review assistance; policy-aware conversational tools (RAG); templated document support; public website search/chat improvements and multilingual access; accessibility remediation; AI-assisted call taking; speech-to-text note-taking with domain awareness; data analysis; translation/transcription; cost-effective access to popular AI tools like Copilot, ChatGPT, Claude; on-premises LLMs and APIs; automation platforms (RPA) and agents; large document search; AI-enabled security/IT operations; AI-assisted development; vision analytics across camera systems; and AI-enhanced ITSM/service desk.
- Functional Domains
•    Forms and Documents
o    Internal and public forms (e.g. forms for services the County provides, interdepartmental forms for business processes) creation and automation with rules-based completeness checks, central electronic form accepting location, prioritization/queuing and automated follow-up.
o    Building plan (drawings, place and other documentation) review assistance using rulesets aligned with code requirements (e.g., building code), including document ingestion and compliance checking.
o    Policy-sourced structured conversation leveraging retrieval-augmented generation (RAG) with citations for internal/external use (e.g., policy libraries, unstructured document search, etc.).
•    Public Website and Resident Services
o    AI-optimized content structure for discoverability (e.g. assisting with Search Engine Optimization for AI); modern multilingual search; and chatbot capabilities that support conversational intake and semi-structured submission.
o    Topic specific chatbots tied to a shared platform architecture for consistent experience (e.g. RAG on a specific documents, policy, or other content that provides more specific information on a topic than a generalized chatbot).
o    Scalable, accurate website translation beyond basic browser tools.
o    Accessibility assessment and automated remediation aligned with web content accessibility guidelines (WCAG); guided fixes for content authors. This should also work on multiple types of content, including websites, PDFs, Images.
o    Non-emergency AI call taking with sentiment detection, multilingual support, and integration to computer-aided dispatch (CAD)/records management system (RMS) and routing systems.
•    Employee Services
o    Speech-to-text notetaking with domain/context awareness; configurable output formats (e.g., medical records or case notes) without diagnostic use.
o    Secure data analysis across regulated datasets; natural language queries; optional onsite operation; and business intelligence (BI) integration.
o    Data discovery, cataloging, classification, and quality services across structured and unstructured sources to make County data “AI‑ready,” including governed data fabric/mesh capabilities for cross‑department access.
o    Translation and transcription services for text, audio, and video with improved workflow efficiency.
o    Cost-effective enterprise access to common large language models (LLMs) (centralized management, usage visibility, guardrails).
o    Enterprise AI platforms and LLM hubs that provide shared APIs/SDKs, connectors, and centralized LLM routing, and cost/usage reporting by department.
o    on-prem LLMs/application programming interfaces (APIs) for sensitive use cases and developer integration (e.g., summarization, ticket title optimization).
o    Automation platforms (robotic process automation) and AI agents for cross system workflows, data cleaning/triage, repetitive task automation, and citizen developer usability.
o    Large, structured/unstructured document search (e.g., FOIA/discovery, IT knowledge bases).
o    Optional agentic and workflow‑centric AI platforms capable of orchestrating multi‑step workflows across County systems (e.g., RPA + LLM + APIs) with human‑in‑the‑loop checkpoints and full action logging.
•    IT-Focused Systems
o    AI Security, Compliance and FOIA Safety: Prompt guardrails, data security/anonymity/exfiltration controls, policy enforcement, logging, misuse detection, FOIA support, eDiscovery support, integration controls, and tool blocking where required.
o    AI governance, guardrails, and compliance tooling aligned with the County’s AI governance framework (e.g., NIST AI RMF), including workflows for use‑case intake, risk assessment, approvals, and automated audit artifacts.
o    Operational Security and IT ops: AI assisted firewall/network/server management, patching, monitoring, trend/log analysis, and data loss prevention.
o    AI-Assisted Development: Enterprise GitHub ecosystem enablement (e.g., Copilot, Advanced Security) with government friendly pricing options and secure code practices.
o    Vision and Cameras: Analytics across diverse camera ecosystems (e.g., Avigilon, FLIR, Axon, Axis, Flock) for smart notifications, recognition, trends, and search. This includes both AI vision enhancements in these tools, as well as tools that can cross various systems.
o    IT Service Management (ITSM) and Service Desk: AI enhanced self-help, intelligent triage/queues, knowledge search, and infrastructure aware alerting.
- AI Infrastructure, Cloud and Cloud Service Provider Requirements
•    AI-optimized computing, storage and networking for on-premises, cloud and edge deployments (e.g., GPU/accelerator platforms, high performance storage, highbandwidth networking).
•    CSP integration and procurement support for major CSPs (e.g., AWS, Azure, Google Cloud, Oracle Cloud), including secure landing zones, connectivity, etc.
•    Architecture and implementation services for AI-capable data center and hybrid cloud environments, including performance tuning and integration with existing backup, disaster recovery, monitoring, and identity platforms.
•    Optional machine learning operations, observability, and model risk management capabilities (model registry, continuous integration/continuous deployment, deployment, monitoring, drift/bias detection, and integration with existing observability/security information and event management tools).