Q: Questions
1. It seems this for chat, where based on eebsite visitor query chat agent pulls relevant answer from knowledge base. Is this the only use if this application?
2. What are the use cases, it can solve?
3. Single vs multi agent in this context. What are underlying LLMs, is it more like wrapper where user interface is integrated with underlying LLM and that's all
4. How can I correlate and understand it in reference to small language model, can you please explain?
5. How are credits being used?
SeanP_AgenticFlowAI
Apr 23, 2025A: Hey VK, happy to clarify:
1. Beyond chat, AgenticFlow is a full no-code automation platform. You can build workflows that pull data from websites, Google Docs, Slack, email, WhatsApp, CRM systems—any of our 2,500+ MCP integrations—and then take actions automatically.
2. Common use cases include customer support bots on your site, marketing automations (social posts, email campaigns), data scraping and reporting, SOP and document generation, video and image creation, P&L analysis, and on-demand research—really anything you can connect via API.
3. A single agent is one AI “employee” that uses tools you equip it with to finish a task. A multi-agent system adds a manager agent that delegates sub-tasks to specialized sub-agents in parallel, making complex, stateful processes more reliable and performant.
4. Under the hood we’re a wrapper around LLM providers and fulfill their missing (give them tools via 2500 MCPs, unified knowledge base). We offer built-in economical models (Gemini Flash, Claude Haiku, etc.) that run on credits, or you can BYOK—plug in your own OpenAI, Anthropic, Google, Perplexity keys or use our OpenRouter pass-through.
5. Credits power all built-in steps: each workflow run charges a small fixed fee plus variable compute credits. If you BYOK, those calls bypass credits and bill directly to your own API account.
Hope that helps!