Shushant Lakhyani

I just enhanced my image in DoodleFusion Overlay style.

Here's how you can do it too.

Use OpenAI GPT Image 2 or any other powerful AI image generation model.

Add your image or the reference image.

Use this prompt: Refine and enhance the uploaded image while strictly preserving the subject’s identity, pose, composition, and lighting. Overlay playful, hand-drawn doodles that intelligently interact with the subject by tracing outlines, exaggerating motion, extending forms, or adding imaginative visual elements that feel responsive to the scene. Use a loose, sketchy, imperfect line style with organic strokes and a natural, hand-illustrated feel. Integrate witty, context-aware handwritten text that complements the mood without relying on generic phrases. Ensure the doodles blend seamlessly into the image with balanced placement, clean layering, and harmonious colors so they enhance rather than overpower the subject. Maintain a vibrant, expressive, and social-media-ready aesthetic in high resolution. The output image must have 1:1 aspect ratio.

I'm sad to see most people still don't know how an AI agent works.

Here's everything about how an AI agent works in an easy-to-understand way.

🔖 Save this post for later

---

1. You give the agent a goal, not a set of rigid instructions.

2. The agent reads your input and breaks the goal into smaller, manageable steps.

3. A large language model (LLM) acts as the brain that does all the thinking and reasoning.

4. The agent builds a plan before it takes any action.

5. It then decides which tools it needs to execute that plan.

6. Tools can include web search, APIs, code runners, databases, or file systems.

7. The agent calls the right tool, reads the output, and moves to the next step.

8. It can loop back and revise its plan if a tool returns an unexpected result.

9. Short-term memory holds the current task context inside the active session.

10. Long-term memory uses vector databases to store and retrieve information across sessions.

11. Retrieval-Augmented Generation (RAG) lets the agent pull relevant facts from stored data.

12. The agent reasons over the retrieved data before writing its final response.

13. Model Context Protocol (MCP) connects agents to external tools and data sources in a standardized way.

14. Agent2Agent protocol (launched by Google in 2025) lets multiple agents talk directly to each other.

15. In a multi-agent system, one orchestrator agent delegates subtasks to specialized sub-agents.

16. Each sub-agent focuses on one job, like writing, searching, or analyzing data.

17. Agents operate almost entirely through APIs and scripts, not visual interfaces.

18. They run fast because they bypass human-style navigation entirely.

19. The agent checks its own output after every action to verify it is on track.

20. If an action fails, the agent retries or picks an alternative approach automatically.

21. Human-in-the-loop checkpoints pause the agent and ask for your approval before risky actions.

22. Guardrails are baked into the system to block the agent from taking harmful or unauthorized steps.

---

To know the next steps,
Check the infographic below 👇

---

Discover the best tools here: https://lnkd.in/daKFh4Gr

---

🔁 Repost to help your network

---

Goodbye manual work.

I don’t understand why people don’t use Google’s AI tools at work.

Here are 22 AI tools by Google to help you get more done in less time.

🔖 Save this post for later

---

Here are all the Google AI tools you’ll get in the carousel below:

1. NotebookLM
2. Gemini
3. Nano Banana Pro
4. Canvas
5. Veo 3.1
6. Guided Learning
7. Visual Layout
8. Gems in Gemini
9. Google AI Studio
10. Deep Research
11. Imagen-4
12. Gemini in Workspace apps
13. AI mode
14. Pomelli
15. Jules
16. Antigravity
17. Opal
18. Vids
19. Flow
20. Stitch
21. Whisk
22. MusicFX

---

Have you tried any of these Google AI tools?

Let me know in the comments 👇

---

Discover the best tools here: https://lnkd.in/daKFh4Gr

---

🔁 Repost to help your network

---

I'm sad to see most people still use Excel and Powerpoint the ancient way.

Here's a detailed guide on how you can use Claude for Excel and Powerpoint to get more done in less time.

🔖 Save this post for later

---

WHAT IS IT?

• Claude works directly inside Microsoft Excel and PowerPoint as official add-ins.
• No tab-switching. No copy-pasting.
• Available on Pro, Max, Team, and Enterprise plans for Mac and Windows.

---

HOW TO INSTALL?

→ Search “Claude by Anthropic for Excel” on Microsoft Marketplace and click Get it now
→ Search “Claude by Anthropic for PowerPoint” and do the same
→ Open each app, activate the add-in from the Home ribbon, and sign in with your Claude account

---

WHAT CLAUDE DOES IN EXCEL?

→ Reads complex multi-tab workbooks and explains formulas with cell-level citations
→ Debugs errors and audits balance-sheet integrity
→ Builds LBO, DCF, and 3-statement financial models
→ Creates pivot tables and charts from your data
→ Updates assumptions without breaking formula dependencies
→ Logs every action in a dedicated “Claude Log” tab

---

To know more,

Check the infographic below 👇

---

Discover the best tools here: https://lnkd.in/daKFh4Gr

---

🔁 Repost to help your network

---