Concepts
Lets get you started with some basic concepts of the platform
Prompts
Introduction
If you have used Chat GPT, you'll be very comfortable with prompting. A prompt is an instruction passed to an LLM, usually in conversational English, which are instructions for the model to act on. Prompts can differ in complexity based on the task's complexity and the LLM that one is utilising to do the task itself.
Our interface doesn't differ much and retains the central text input field for you to put in your prompt and save it. We have added some brilliant features which will help you create uber-powerful agents, which you can check how to use here.
These prompts can be further fine-tuned (coming soon 😁) and improved by changing the model parameters like temperature, max tokens, etc, to get the best and desired outcome.
Prompt Engineering
Prompt engineering is the process of optimizing your prompts through repeated testing. This can be done either through trial and error (just rewrite prompts and check the output), or programmatically (generate prompts, then test, and use another LLM to evaluate the responses).
For comprehensive guidance on advanced prompting techniques and best practices, refer to these detailed documentation resources: OpenAI Prompting Guide & Anthropic Claude 4 Best Practices.
Although we don't support prompt engineering infrastructure, you can still do basic prompt engineering by trial-and-error testing.
Agents
AI agents are an encapsulation around language models which help you do advanced tasks since they have the following capabilities:
Memory persistence: Agents can retain previously done tasks, actions or interactions, therefore, they can maintain context in real-time.
Autonomous actions: If instructed accordingly, agents can take actions like calling API tools, searching the web, referring databases, etc.
Self-reasoning: They can reflect on their outputs and further build on their previous reasoning if requested to do the same in their instructions.
Self-improving: Agents can retrospectively examine their previous responses and use this information and feedback to improve their future output.
Our platform empowers you to build and deploy multiple powerful AI agents for your business operations, allowing you to automate intra- and inter-organizational tasks across all verticals like sales, marketing, customer support, and logistics.
Flows
Flows are decision paths that one can deploy to connect multiple agents and tools to automate subjective and objective tasks. Flows are the main medium to build any kind of automation on the platform, be it conversational LLMs, scheduled data egress/ingress, agentic data enrichment or platform actions like updating tickets.
We will discuss flows in depth under the basics of flows here.
You can also directly jump to the flow tutorials here.
Forms/Attributes
The Forms section enables teams to create structured data collection templates that integrate directly with AI prompts, task flows, and customer workflows. These forms act as reusable input modules that standardize and capture user data effectively for personalized automation.
Tools
The Tools section enables you to create specialized utilities that extend your platform's capabilities through AI-powered processing and external API integrations, providing custom functionalities for enhanced workflow automation.
Knowledge Base
The Knowledge Bases section under the Data module provides centralized content management for powering AI agents, automated responses, and customer support workflows. This system enables teams to organize, version, and deploy structured information across all customer touchpoints.
Knowledge Reference
Knowledge Base References create targeted subsets of your knowledge base content that can be called and referenced throughout angoor.ai. Think of them as curated collections that combine specific knowledge bases into focused, reusable resources for different contexts and use cases.
Last updated