Insights

Substack AI Studio

2025-11-28

Stop Spending on Studios: 10 Prompts for the Perfect LinkedIn Photo

Practical tutorial to create a professional LinkedIn profile photo using Google AI Studio (Build) and image editing with the 'Nano Banana' model to preserve your features. Includes step-by-step (selfie upload, prompt, iterations) and 10 ready-made prompts for executive, tech, and creative styles, focusing on realism and good lighting.

• Edit a selfie (don't generate from scratch) to maintain identity and consistency.
• In the prompt, specify lens/lighting, background, clothing, framing, and constraints (no filters, no caricature).
• Use the 10 templates as a base and refine in short iterations until it looks natural.

Read on Substack

LinkedIn AI Agents

2026-02-04

Claude Skills No-Code: Installing Skills via Manus AI

Quick guide to using Claude Skills without writing code: import a skill from GitHub within Manus AI and run commands like /seo-audit. The post shows skill repositories, the critical tip of copying the folder URL with SKILL.md, and the import flow (Manus -> + -> Skills -> Add from GitHub).

• Choose the right skill and copy the folder URL (not the root repo).
• Install in minutes via interface and reuse as commands in chat.
• Example: run an SEO audit targeted at the audience.

View on LinkedIn

LinkedIn Privacy

2026-01-19

Confer: Privacy-Focused ChatGPT Alternative

Introduces Confer (confer.to), a ChatGPT alternative focused on privacy created by Signal's co-founder. The differentiator is a key/passkey generated at registration (stored in password vault) to encrypt conversations. Cites initial price (US$35) and use of open source LLMs without specifying which.

• Privacy as a product: local key + encryption (according to the proposal).
• Trade-off: price above competitors and limited transparency about models.
• Option to evaluate in sensitive flows (legal, compliance, internal data).

View on LinkedIn

LinkedIn Leadership

2026-01-11

Linus, AI, and Leverage: Efficiency Beats Purism

A real phrase from a Linus Torvalds commit becomes an argument about AI as leverage: in specific tasks (e.g., visualization), the machine can do better than 'by hand'. The message for leadership is pragmatism: it's not about replacing expertise, it's about increasing throughput and quality when it makes sense.

• Frame AI as a leverage tool in delimited tasks.
• Create culture to admit 'better than by hand' without ego.
• Define criteria: where AI improves result vs where risk doesn't compensate.

View on LinkedIn

LinkedIn Tech Trends

2026-01-01

What to Expect for Technology in 2026 (NotebookLM + Deep Research)

Study compiled in NotebookLM with sources from consultancies and big techs, supported by Deep Research. Synthesis in 4 points: AI becomes 'invisible' and ubiquitous, autonomous agents make business decisions, human skills become differentiator, and AI infrastructure cost grows and starts being charged.

• Prepare governance for agents (risk, audit, accountability).
• Invest in human skills (critical thinking, communication, leadership).
• Plan FinOps/chargeback for AI: inference/training cost enters P&L.

View on LinkedIn

LinkedIn AI Agents

2025-12-29

Meta and Manus AI: Agents with VM, Research, and Deliverables

Note on Meta's investment/acquisition in Manus AI, an agent platform that executes tasks via virtual machines, conducts research, and generates deliverables (e.g., PPT). Based on usage, the post suggests the investment accelerates maturity and adoption.

• Agents with real execution (VMs) expand automation beyond chat.
• Evaluate use cases: research, asset generation, repetitive operational routines.
• Define guardrails: permissions, human review, and limits for critical tasks.

View on LinkedIn

LinkedIn Automation

2025-10-06

OpenAI's Agent Builder: End of n8n or Market Expansion?

Reaction to OpenAI's Agent Builder and the 'panic' that Zapier/n8n/Make and automation agencies would end. The argument: much (estimated ~60%) of automation is still traditional (integrations, APIs, rules), and Agent Builder tends to expand the market and bring more visibility to low-code platforms.

• Separate deterministic automation from probabilistic steps with AI.
• Agent Builder can be an entry point; low-code platforms remain essential.
• Map flows: where AI adds value vs where pure integration solves.

View on LinkedIn

LinkedIn Decision Making

2025-09-26

Turning Back Is Not Weakness: Integrity in Decision Making

Reflection on decision making: turning back is not weakness, but integrity when values and facts change. Highlights courage to recognize unfeasible commitments, take responsibility with transparency, prioritize values over vanity, and correct decisions influenced by biases. 'Turning back is also a decision'.

• Design reversibility: decision records, checkpoints, and kill switches.
• Normalize review: changing mind with data is a sign of maturity.
• Transparency + repair sustain trust more than unrealistic promises.

View on LinkedIn

LinkedIn Prompt Engineering

2025-08-29

Prompt Engineering: Role, Task, Audience, Tone, Constraints

Video and text explaining why generic answers come from generic prompts. Recommends treating prompts as instructions: define Role, Task, Audience, Tone, and Format/Constraints. Emphasizes iteration: the first result is a draft, you direct the assistant.

• Role + Task align expertise and expected deliverable.
• Audience + Tone adjust depth and style for context.
• Constraints reduce verbosity and keep output in the right format.

View on LinkedIn

LinkedIn NotebookLM

2025-08-28

NotebookLM: Sources by Link (YouTube/Sites) and Video Summary

Google NotebookLM update: beyond PDFs/text/audio, now accepts website and YouTube links as sources, and generates summaries not only in audio (podcast) but also in video. The post shows a test loading 4 talks about Context Engineering and requesting a video summary.

• Source curation becomes pipeline: collect links and 'converse' with the dataset.
• Video summary accelerates alignment with team and stakeholders.
• Use for research, onboarding, and technical review with traceability.

View on LinkedIn

LinkedIn Context Engineering

2025-08-23

Vibe Coding + Context Engineering: How to Move Beyond Loose Prompts

Vibe Coding is programming with AI in exploratory, fast, flow mode. The quality leap comes with Context Engineering: organize objectives, constraints, examples, history, and references for the AI to understand the problem and stop responding generically. Points to resources like YouTube, LangChain, BMAD Method, and Super Claude.

• Without context: more trial and error; with context: speed with quality.
• Standardize artifacts (living docs, story files) to maintain project memory.
• Frameworks (LangChain/BMAD/Super Claude) help operationalize the method.

View on LinkedIn

LinkedIn AI Adoption

2025-08-01

Generative AI with Common Sense: When RPA Is Better

Adopting generative AI with common sense: not everything needs to be automated with AI. For some tasks, human touch is essential; for others, traditional RPA may be faster and cheaper. The point is to decide where AI really adds value vs where it only adds cost and complexity.

• Use criteria: value, risk, cost, latency, and need for explainability.
• Combine AI + RPA when it makes sense (AI classifies, RPA executes).
• Keep human in the loop in sensitive decisions and external communications.

View on LinkedIn

LinkedIn AI Agents

2025-07-24

Manus AI Introduces Itself: Business and Nerd in the Same Agent

Experiment asking Manus AI to introduce itself, highlighting the platform's versatility: 'business and nerd' at the same time, delivering relevant content and code. The post mentions ongoing tests and a link/code to earn more credits.

• Use self-introduction prompts to understand capabilities and limits.
• Validate in 2 dimensions: executive output (business) and technical (code).
• Start with small, real tasks before scaling.

View on LinkedIn

LinkedIn Visualization

2025-07-19

Napkin AI: Transform Text into Visualizations in 1 Click

Recommendation of Napkin AI (app.napkin.ai) to quickly transform text into graphical visualizations. Flow: generate or upload content, select excerpts and trigger the 'lightning' button to get various visual interpretations that summarize or detail the idea. Focus on productivity for communication and alignment.

• Good for transforming drafts into diagrams/visuals without manual design.
• Use in architecture, strategy, roadmaps, and pitch decks.
• Iterate quickly: compare visualizations and choose the clearest.

View on LinkedIn

LinkedIn Automation

2025-07-18

n8n with AI Nodes: Workflows That Think Before Acting

How n8n evolves with native AI nodes: flows connect WhatsApp, databases, audio transcription, enrichment and summary, generating dynamic responses with context. Automation ceases to be just 'if/then' and starts to 'think' (exceptions, writing, classification, translation), integrating OpenAI or Gemini without separate backend.

• Architecture: low-code orchestration + LLM as workflow step (no glue code).
• Use AI where there's ambiguity; keep deterministic in the rest of the flow.
• Governance: logs, prompt versioning, cost limits, and fallback.

View on LinkedIn

LinkedIn Security

2023-01-05

Flipper Zero: Banning May Popularize and Create Clones

Reflection on Flipper Zero: similar old devices exist in parallel markets; Flipper's advantage would be gathering functions and having an intuitive UX to explore communication patterns without much study. The prediction is that prohibitions may have the opposite effect, increasing popularity and stimulating clones.

• Security by prohibition usually fails; reduce risk with education and controls.
• Hardware accessibility tends to increase: prepare policies and detection.
• For defense, focus on hardening and monitoring protocols/radio.

View on LinkedIn