Back

Rahul Agarwal

Founder | Agentic AI...ย โ€ขย 20d

4 ways how AI systems communicate and coordinate. I've explained each one in detail below. 1. ๐— ๐—–๐—ฃ (๐— ๐—ผ๐—ฑ๐—ฒ๐—น ๐—–๐—ผ๐—ป๐˜๐—ฒ๐˜…๐˜ ๐—ฃ๐—ฟ๐—ผ๐˜๐—ผ๐—ฐ๐—ผ๐—น) โ€ข User submits a request: โ€œSummarize todayโ€™s Slack messages.โ€ โ€ข MCP Client receives input: Interface between user and AI system. โ€ข Request forwarded to MCP Host: Host manages execution and control. โ€ข Model selection: MCP Host decides which LLM handles request. โ€ข Tool permission check: MCP Host defines allowed tools (Slack, DBs). โ€ข LLM processes intent: Understands task and decides if tool needed. โ€ข MCP Server executes tool calls: Communicates with Slack, Qdrant, Brave. โ€ข Tool data returned: Real-world data flows back to model. โ€ข Final response generated: Tool โ†’ LLM โ†’ MCP Client โ†’ User. ๐— ๐—–๐—ฃ = ๐—ฆ๐—ฎ๐—ณ๐—ฒ, ๐˜€๐˜๐—ฟ๐˜‚๐—ฐ๐˜๐˜‚๐—ฟ๐—ฒ๐—ฑ ๐˜„๐—ฎ๐˜† ๐—ณ๐—ผ๐—ฟ ๐—Ÿ๐—Ÿ๐— ๐˜€ ๐˜๐—ผ ๐˜‚๐˜€๐—ฒ ๐˜๐—ผ๐—ผ๐—น๐˜€ _____________ 2. ๐—”2๐—” (๐—”๐—ด๐—ฒ๐—ป๐˜-๐˜๐—ผ-๐—”๐—ด๐—ฒ๐—ป๐˜ ๐—ฃ๐—ฟ๐—ผ๐˜๐—ผ๐—ฐ๐—ผ๐—น) โ€ข User gives a complex task: โ€œResearch competitors and summarize insights.โ€ โ€ข Primary agent receives task: Acts as coordinator. โ€ข Agent discovery begins: System identifies agents with relevant skills. โ€ข A2A protocol establishes communication: Defines how agents exchange messages. โ€ข Task delegation happens: Research, analysis, summary agents. โ€ข Agents work independently: Each may use different models or tools. โ€ข Progress shared between agents: Partial results stream back. โ€ข Results aggregated: Primary agent combines outputs. โ€ข Final response delivered. ๐—”2๐—” = ๐—”๐—œ ๐—ฎ๐—ด๐—ฒ๐—ป๐˜๐˜€ ๐—ฐ๐—ผ๐—น๐—น๐—ฎ๐—ฏ๐—ผ๐—ฟ๐—ฎ๐˜๐—ถ๐—ป๐—ด ๐—น๐—ถ๐—ธ๐—ฒ ๐˜๐—ฒ๐—ฎ๐—บ๐—บ๐—ฎ๐˜๐—ฒ๐˜€ ____________ 3. ๐—”๐—–๐—ฃ (๐—”๐—ด๐—ฒ๐—ป๐˜ ๐—–๐—ผ๐—บ๐—บ๐˜‚๐—ป๐—ถ๐—ฐ๐—ฎ๐˜๐—ถ๐—ผ๐—ป ๐—ฃ๐—ฟ๐—ผ๐˜๐—ผ๐—ฐ๐—ผ๐—น) โ€ข User sends request via ACP Client: Entry point to agent system. โ€ข ACP Client routes request to Agent 1: First responding agent. โ€ข Agent exposes profile: Capabilities, tools, limits. โ€ข ACP defines message structure: Standard metadata and formats. โ€ข Agent discovery occurs: Other compatible agents identified. โ€ข Second agent joins: May use different tools or frameworks. โ€ข Protocol ensures understanding: All agents speak same โ€œlanguageโ€. โ€ข Structured communication happens: No custom formats or assumptions. โ€ข Response flows back: Clean and predictable output. ๐—”๐—–๐—ฃ = ๐—ฃ๐—ฟ๐—ผ๐—ณ๐—ถ๐—น๐—ฒ + ๐—ต๐—ฎ๐—ป๐—ฑ๐˜€๐—ต๐—ฎ๐—ธ๐—ฒ ๐˜€๐˜๐—ฎ๐—ป๐—ฑ๐—ฎ๐—ฟ๐—ฑ ๐—ณ๐—ผ๐—ฟ ๐—ฎ๐—ด๐—ฒ๐—ป๐˜๐˜€ ____________ 4. ๐—”๐—ก๐—ฃ (๐—”๐—ด๐—ฒ๐—ป๐˜ ๐—ก๐—ฒ๐˜๐˜„๐—ผ๐—ฟ๐—ธ ๐—ฃ๐—ฟ๐—ผ๐˜๐—ผ๐—ฐ๐—ผ๐—น) โ€ข User submits a real-world request: โ€œPlan a complete trip.โ€ โ€ข Coordinator agent activated: Understands overall goal. โ€ข Relevant agents identified: Travel, hotel, weather agents. โ€ข ANP manages interactions: Routing requests and responses. โ€ข Agents share partial results: Availability, prices, forecasts. โ€ข Dependencies resolved: One agentโ€™s output feeds another. โ€ข Network stays synchronized: Information remains consistent. โ€ข Final solution assembled: End-to-end plan created. โ€ข User receives outcome. ๐—”๐—ก๐—ฃ = ๐— ๐˜‚๐—น๐˜๐—ถ-๐—ฎ๐—ด๐—ฒ๐—ป๐˜ ๐˜๐—ฒ๐—ฎ๐—บ๐˜„๐—ผ๐—ฟ๐—ธ ๐—ณ๐—ผ๐—ฟ ๐—ฐ๐—ผ๐—บ๐—ฝ๐—น๐—ฒ๐˜… ๐˜„๐—ผ๐—ฟ๐—ธ๐—ณ๐—น๐—ผ๐˜„๐˜€ โœ… Repost for other people in your network so they can understand this.

1 Reply
5
Replies (1)

More like this

Recommendations from Medial

Image Description

Rahul Agarwal

Founder | Agentic AI...ย โ€ขย 1m

2 main frameworks powering todayโ€™s AI workflows. Iโ€™ve explained both in simple steps below. ๐—ก๐Ÿด๐—ก (๐—Ÿ๐—ถ๐—ป๐—ฒ๐—ฎ๐—ฟ ๐—”๐—ด๐—ฒ๐—ป๐˜ ๐—™๐—น๐—ผ๐˜„) (๐˜ด๐˜ต๐˜ฆ๐˜ฑ-๐˜ฃ๐˜บ-๐˜ด๐˜ต๐˜ฆ๐˜ฑ) N8N lets AI follow a ๐˜€๐˜๐—ฟ๐—ฎ๐—ถ๐—ด๐—ต๐˜, ๐˜ƒ๐—ถ๐˜€๐˜‚๐—ฎ๐—น ๐˜„๐—ผ๐—ฟ๐—ธ๐—ณ๐—น๐—ผ๐˜„, moving step-by-step

See More
1 Reply
6

Rahul Agarwal

Founder | Agentic AI...ย โ€ขย 18d

Most people even today don't know this about MCP. I've explained it in simple way below. AI systems fail because control logic lives inside prompts. MCP moves that control outside the model, where it belongs. 1. ๐——๐—ถ๐—ฟ๐—ฒ๐—ฐ๐˜ ๐—”๐—ฃ๐—œ ๐—ช๐—ฟ๐—ฎ๐—ฝ๐—ฝ๐—ฒ๐—ฟ

See More
Reply
1

Rahul Agarwal

Founder | Agentic AI...ย โ€ขย 2m

MCP is getting attention, but itโ€™s just one piece of the puzzle If youโ€™re developing Agentic AI systems, itโ€™s crucial to understand more than just MCP. There are 5 key protocols shaping how AI agents communicate, collaborate, and scale intelligence

See More
Reply
8

Rahul Agarwal

Founder | Agentic AI...ย โ€ขย 3m

3 core protocols behind AI agents. Iโ€™ve explained each one in simple steps. 1. ๐— ๐—–๐—ฃ (๐— ๐—ผ๐—ฑ๐—ฒ๐—น ๐—–๐—ผ๐—ป๐˜๐—ฒ๐˜…๐˜ ๐—ฃ๐—ฟ๐—ผ๐˜๐—ผ๐—ฐ๐—ผ๐—น) โ€ข ๐— ๐—ฎ๐—ถ๐—ป ๐—ฃ๐˜‚๐—ฟ๐—ฝ๐—ผ๐˜€๐—ฒ: It helps AI models (LLMs) use extra information by connecting them to tools or databases.

See More
Reply
3
5
Image Description

Rahul Agarwal

Founder | Agentic AI...ย โ€ขย 4m

How Multi-Agent AI systems actually work? Explained in a very simple way. Read below: -> ๐—ง๐—ต๐—ฒ ๐— ๐—ฎ๐—ถ๐—ป ๐—”๐—œ ๐—”๐—ด๐—ฒ๐—ป๐˜ The main ๐—”๐—œ ๐—”๐—ด๐—ฒ๐—ป๐˜ is the ๐—ผ๐—ฟ๐—ฐ๐—ต๐—ฒ๐˜€๐˜๐—ฟ๐—ฎ๐˜๐—ผ๐—ฟ. It has several capabilities: โ€ข ๐——๐—ฎ๐˜๐—ฎ๐—ฏ๐—ฎ๐˜€๐—ฒ โ€“ Stores knowledge o

See More
Reply
6
19
1
Image Description

Rahul Agarwal

Founder | Agentic AI...ย โ€ขย 1m

4 core ways multi-agent AI systems are designed. Iโ€™ve explained each one in simple steps below. 1. ๐—ฃ๐—ฎ๐—ฟ๐—ฎ๐—น๐—น๐—ฒ๐—น ๐—ฃ๐—ฎ๐˜๐˜๐—ฒ๐—ฟ๐—ป (๐˜š๐˜ต๐˜ฆ๐˜ฑ-๐˜ฃ๐˜บ-๐˜ด๐˜ต๐˜ฆ๐˜ฑ) โ€ข One input (task) comes in. โ€ข The task is ๐˜€๐—ฝ๐—น๐—ถ๐˜ ๐—ถ๐—ป๐˜๐—ผ ๐—ฝ๐—ฎ๐—ฟ๐˜๐˜€. โ€ข Multiple AI age

See More
Reply
4
10
1

Rahul Agarwal

Founder | Agentic AI...ย โ€ขย 1m

Hands down the simplest explanation of AI agents using LLMs, memory, and tools. A user sends an input โ†’ the system (agent) builds a prompt and may call tools and memory-search (RAG) โ†’ agent decides and builds an answer โ†’ the answer is returned to th

See More
Reply
2
7
Image Description

Rahul Agarwal

Founder | Agentic AI...ย โ€ขย 10d

Most people overlook these basics of AI Agents. I've explained it in a very simple way below. 1. ๐—”๐—œ ๐—”๐—ด๐—ฒ๐—ป๐˜ An AI system that observes its environment, information, makes decisions, and takes actions to achieve a goal. 2. ๐—Ÿ๐—Ÿ๐— ๐˜€ (๐—Ÿ๐—ฎ๐—ฟ๐—ด๐—ฒ

See More
Reply
6
1

Rahul Agarwal

Founder | Agentic AI...ย โ€ขย 15d

Most people building modern AI systems miss these steps. I've explained each step in a simple way below. 1. ๐— ๐˜‚๐—น๐˜๐—ถ-๐—”๐—ด๐—ฒ๐—ป๐˜ ๐—œ๐—ป๐˜๐—ฒ๐—ฟ๐—ผ๐—ฝ๐—ฒ๐—ฟ๐—ฎ๐—ฏ๐—ถ๐—น๐—ถ๐˜๐˜† How multiple AI agents work together as a system. Step-by-step: โ€ข ๐—จ๐˜€๐—ฒ๐—ฟ ๐—ฅ๐—ฒ๐—พ๐˜‚๏ฟฝ

See More
2 Replies
1
8

Download the medial app to read full posts, comements and news.