AI Service Activities
The AI Service Activities package enables Akabot robots to integrate seamlessly with leading Generative AI and Large Language Model (LLM) providers. By bringing state-of-the-art cognitive capabilities directly into your automated workflows, you can automate complex decision-making, synthesize unstructured data, categorize emails, extract key details from documents, and drive natural conversation loops.
Supported AI Providers
This package is provider-agnostic, allowing you to connect to and transition between multiple top-tier AI services with minimal configuration:
| Provider | Supported Models (Examples) | Typical Use Cases |
|---|---|---|
| OpenAI | gpt-4o, gpt-4, gpt-3.5-turbo | General reasoning, structured data extraction, agentic tasks. |
| Google Gemini | gemini-1.5-pro, gemini-1.5-flash | High-context window tasks, rapid text generation, multimodal processing. |
| Anthropic Claude | claude-3-5-sonnet, claude-3-opus, claude-3-haiku | Detailed drafting, advanced reasoning, complex coding/logic assistance. |
| Azure OpenAI | Enterprise-grade deployments of OpenAI models | Secure, enterprise-managed environments with strict data privacy. |
Package Key Features
- Provider-Agnostic Container (
AI Scope): Manage credentials, connections, default timeouts, and session states in a single parent scope. Child activities inside automatically inherit the established connection. - Efficient Session Reusability: Share active connection sessions across multiple scopes in a workflow to minimize authorization overhead and latency.
- Text Completion: Generate quick responses, complete prompt-based tasks, summarize lengthy documents, or format unstructured information.
- Chat Completion: Engage in conversational multi-turn flows, apply system-level guidelines, handle high-volume interactions, and configure parameters such as Temperature, Max Tokens, and Stop Sequences.
Activity Catalog
The package includes the following activities categorized by function:
1. Connection & Scope
- AI Scope: Authenticates with the selected provider (OpenAI, Gemini, Anthropic, or Azure OpenAI) using your API Key and Model configurations, serving as the execution scope for all other AI activities.
2. Chat Completion
These activities let you build interactive conversational workflows using standard chat endpoints:
- Generate Chat Completion: Drives chat interactions using OpenAI models.
- Generate Chat Completion Using Gemini: Drives chat interactions using Google Gemini models.
- Generate Chat Completion Using Anthropic: Drives chat interactions using Anthropic Claude models.
- Generate Chat Response Azure OpenAI: Drives chat interactions using Azure OpenAI hosted models.
3. Text Completion
These activities generate singular completions or handle prompt-and-response tasks:
- Generate Text Completion: Performs text completion using OpenAI models.
- Generate Text Completion Using Gemini: Performs text completion using Google Gemini models.
- Generate Text Completion Using Anthropic: Performs text completion using Anthropic Claude models.
- Generate Text Response Azure OpenAI: Performs text completion using Azure OpenAI hosted models.
How to Get Started
Building a workflow with AI Service Activities follows a simple, consistent pattern:
- Add the Container: Drag and drop the AI Scope activity into your workflow designer.
- Setup Authentication: In the AI Scope properties, choose your Provider Type, paste your API Key, and specify the desired Model (e.g.,
gpt-4oorgemini-1.5-flash). - Insert an AI Activity: Place your desired text or chat activity (e.g., Generate Chat Completion) directly inside the Do block of the AI Scope.
- Map Prompt and Output: Write your input prompt in the activity's body. Create a variable in the Result field (under the Output properties) to capture the AI's generated response for downstream processing.