AI Workflow Enhancement

I think AI agentic workflows will drive massive AI progress this year — perhaps even more than the next generation of foundation models. This is an important trend, and I urge everyone who works in AI to pay attention to it. Today, we mostly use LLMs in zero-shot mode, prompting a model to generate final output token by token without revising its work. This is akin to asking someone to compose an essay from start to finish, typing straight through with no backspacing allowed, and expecting a high-quality result. Despite the difficulty, LLMs do amazingly well at this task! With an agentic workflow, however, we can ask the LLM to iterate over a document many times. For example, it might take a sequence of steps such as: - Plan an outline. - Decide what, if any, web searches are needed to gather more information. - Write a first draft. - Read over the first draft to spot unjustified arguments or extraneous information. - Revise the draft taking into account any weaknesses spotted. - And so on. This iterative process is critical for most human writers to write good text. With AI, such an iterative workflow yields much better results than writing in a single pass. Devin’s splashy demo recently received a lot of social media buzz. My team has been closely following the evolution of AI that writes code. We analyzed results from a number of research teams, focusing on an algorithm’s ability to do well on the widely used HumanEval coding benchmark. You can see our findings in the diagram below. GPT-3.5 (zero shot) was 48.1% correct. GPT-4 (zero shot) does better at 67.0%. However, the improvement from GPT-3.5 to GPT-4 is dwarfed by incorporating an iterative agent workflow. Indeed, wrapped in an agent loop, GPT-3.5 achieves up to 95.1%. Open source agent tools and the academic literature on agents are proliferating, making this an exciting time but also a confusing one. To help put this work into perspective, I’d like to share a framework for categorizing design patterns for building agents. My team AI Fund is successfully using these patterns in many applications, and I hope you find them useful. - Reflection: The LLM examines its own work to come up with ways to improve it. - Tool use: The LLM is given tools such as web search, code execution, or any other function to help it gather information, take action, or process data. - Planning: The LLM comes up with, and executes, a multistep plan to achieve a goal (for example, writing an outline for an essay, then doing online research, then writing a draft, and so on). - Multi-agent collaboration: More than one AI agent work together, splitting up tasks and discussing and debating ideas, to come up with better solutions than a single agent would. I’ll elaborate on these design patterns and offer suggested readings for each next week. [Original text: https://lnkd.in/gSFBby4q ]

I see many people struggling or confused when switching into AI. Don’t jump straight into frameworks like LangChain or LangGraph. Frameworks are accelerators, not starting points. Without foundations, you’ll end up building fragile demos instead of production-grade systems. Here’s a step-by-step path to transition your career into Generative AI: 1. Build Core Foundations --Python (APIs, JSON, virtual envs, packaging) --Git, Docker, Linux basics --Databases: Postgres + pgvector, or FAISS for embeddings 2. Learn Just Enough Math & Data --Vectors, cosine similarity, probability --Tokenization, chunking, normalization 3. Understand LLM Basics --How transformers work at a high level --Different types of models: base vs. instruct, hosted vs. local --Prompt engineering patterns (instruction, few-shot, tool-use) 4. Get Hands-on with RAG (without frameworks first) --Ingest → chunk → embed → store → retrieve → re-rank → generate --Add logging, caching, retries --Evaluate outputs with ground-truth sets 5. Learn Evaluation & Safety --Handle hallucination, PII, toxicity --Define and track metrics (accuracy, latency, cost) 6. Explore Reliability & MLOps --CI/CD for prompts/config --Observability, tracing, cost dashboards --Error handling and fallbacks 7. Then Explore Agents --Start simple: one-tool agents --Add planning and memory only when metrics prove value 8. Finally → Use Frameworks Wisely --Adopt LangChain, LangGraph, or LlamaIndex as orchestration layers --Keep your core logic framework-agnostic 9. Showcase Projects --Document QA system with metrics --Structured extraction pipeline with redaction --A small but reliable agent automating a real workflow 10. Be Interview-Ready --Explain RAG pipelines on a whiteboard --Compare models and providers --Justify design choices (chunking, caching, re-ranking) Learn the primitives first. Frameworks make you faster after you understand what’s under the hood. That’s how you build systems that last.

Just built an IT Audit AI agent use case — and it’s only the beginning. Having shared posts over the past few weeks about the disruption Agentic AI is bringing to technology auditing, I thought… why not get stuck in myself? What many people don’t know: I started out as a coder. My early career included hacking Linux as a teenager — and now, years later, I’ve rekindled that curiosity. As a leader, I believe it’s important to understand things hands-on — especially when advising clients or coaching teams. I recently highlighted the emerging skills needed in risk and audit, and decided to upgrade myself. This week, I built a document summarisation agent using Python and OpenAI. I know there’s tons out there, but the point here is I put this together myself and any IT Internal Audit/ Compliance team can do this too! It scans long PDF reports, extracts key insights, and generates audit-relevant summaries — within seconds. Think about the time saved on policies, SOC reports, or control narratives. But more than that: Agentic AI isn’t just a trend. It’s a smart, practical partner for IT risk professionals. I’ve written a whitepaper on “Agentic AI in IT Risk & Controls” exploring how goal-driven agents can enable: • Continuous control monitoring • Automated evidence collection • Change risk evaluation in DevOps • Vendor risk review automation • Narrative generation for audit reports Client example: One of our consumer & retail clients manages over 20 third-party systems. To improve access reviews and change assurance, we deployed an agent that connects to ServiceNow and IAM logs, flags risky changes, and drafts evidence summaries. It’s reducing audit prep time and enabling real-time insight. At EY, our audit and technology risk teams are diving deep into how Agentic AI is disrupting business processes. We’re mapping out new risk models and defining what future-ready auditors will need in terms of skills and mindset. We’ll be sharing our views on this changing compliance landscape soon. Big thanks to Piers C. Maree-Louise Kernick Gareth James Abhishek Mohal Jason Walters Danila Solovyev Kevin Duthie — and everyone else helping shape the future of digital trust. If you’re in risk, audit, or compliance — or just curious about how AI is reshaping our field — let’s connect. The journey has only just begun. #ITRisk #AI #AgenticAI #Python #AuditInnovation #OpenAI #RiskTransformation #DigitalTrust #Leadership

Most people drown in the endless sea of new AI tools. But the truth is - you don’t need hundreds of tools to stay ahead in 2026. You only need to master the 10 categories that actually drive business results, automation, and career acceleration. This guide breaks them down with clarity: what you need, why it matters, and the real impact each category delivers. Here’s the snapshot: 🔹 1. Advanced LLMs (Your New Thinking Models) ChatGPT, Claude, Gemini, Llama, DeepSeek → These become your operating system for reasoning, analysis, writing, coding, planning, and problem-solving. 🔹 2. AI Automation Tools (Workflow Builders) Make.com, n8n, Zapier, Pipedream → The backbone of automated sales, onboarding, support, content pipelines, and internal systems. 🔹 3. AI Agents & Orchestration Tools CrewAI, LangChain, LlamaIndex, AutoGen, OpenAI → 2026 is about multi-step workflows and self-correcting agents that function like digital employees. 🔹 4. Vector Databases (Memory for AI Systems) Pinecone, Weaviate, ChromaDB, Milvus → The foundation of RAG applications, internal chatbots, and knowledge automation. 🔹 5. Knowledge Management + Document Intelligence Notion AI, Airtable AI, Secoda, Glean, Elastic AI → Instant summaries, automated documentation, and searchable intelligence hubs for faster decision-making. 🔹 6. AI Video & Avatar Tools Synthesia, HeyGen, Runway, Pika → Training, marketing, and onboarding videos created in minutes - video becomes the default communication layer. 🔹 7. AI Data Tools (Analytics + Insights Engines) ClickUp AI, Tableau AI, PowerBI AI, Amplitude AI, Akkio → Automated dashboards, predictive insights, and analytics without needing SQL or code-heavy workflows. 🔹 8. AI Design Tools (Visual Experience Builders) Canva AI, Adobe Firefly, MidJourney, Figma AI → Branding, ads, UI/UX, infographics, thumbnails - all created 10× faster through prompting. 🔹 9. AI Coding Tools GitHub Copilot, Cursor, Replit AI, Codeium → Faster builds, fewer bugs, and better architecture. Developers shift from code writers to solution architects. 🔹 10. AI Search & Personal Intelligence Tools Perplexity, LexisNexis AI, Adobe Ask → Instant reports, automated research, competitor analysis, and conversational search. This is the real AI stack for 2026. Not hype. Not noise. Just the tools that will genuinely move your business, your work, and your career forward. Which category are you focusing on next?

Google just launched Agent2Agent (A2A) protocol that could quietly reshape how AI systems work together. If you’ve been watching the agent space, you know we’re headed toward a future where agents don’t just respond to prompts. They talk to each other, coordinate, and get things done across platforms. Until now, that kind of multi-agent collaboration has been messy, custom, and hard to scale. A2A is Google’s attempt to fix that. It’s an open standard for letting AI agents communicate across tools, companies, and systems, that securely, asynchronously, and with real-world use cases in mind. What I like about it: - It’s designed for agent-native workflows (no shared memory or tight coupling) - It builds on standards devs already know: HTTP, SSE, JSON-RPC - It supports long-running tasks and real-time updates - Security is baked in from the start - It works across modalities- text, audio, even video But here’s what’s important to understand: A2A is not the same as MCP (Model Context Protocol). They solve different problems. - MCP is about giving a single model everything it needs- context, tools, memory, to do its job well. - A2A is about multiple agents working together. It’s the messaging layer that lets them collaborate, delegate, and orchestrate. Think of MCP as helping one smart model think clearly. A2A helps a team of agents work together, without chaos. Now, A2A is ambitious. It’s not lightweight, and I don’t expect startups to adopt it overnight. This feels built with large enterprise systems in mind, teams building internal networks of agents that need to collaborate securely and reliably. But that’s exactly why it matters. If agents are going to move beyond “cool demo” territory, they need real infrastructure. Protocols like this aren’t flashy, but they’re what make the next era of AI possible. The TL;DR: We’re heading into an agent-first world, and that world needs better pipes. A2A is one of the first serious attempts to build them. Excited to see how this evolves.

Over the past few weeks, I’ve had time to reflect — between AI briefing sessions, leading workshops, and teaching at LinkedIn offices in London. One question keeps coming up: “Will I still have a job?” It’s a fair question. The reports, social posts, and videos can be overwhelming — even frightening. But here’s what I’ve come to believe: -- AI isn’t replacing jobs outright. It’s replacing tasks. And that changes everything. In my role as Course Director at the CIM | The Chartered Institute of Marketing — and through my work with marketers, consultants, and business owners — I’ve seen the same shift again and again: Those who are moving forward are reframing their role. They’re not competing with AI — they’re learning how to work alongside it. Here’s how I break down the Human + AI relationship in practical terms: 👉 20% – Spark Ideas Use AI to generate ideas, explore angles, and overcome creative blocks. 👉 40% – Co-Pilot Mode Let AI support your process — drafting, outlining, and iterating while you steer. 👉 60% – Efficiency Booster AI automates and structures routine work so you can focus on higher-value thinking. 👉 80% – Heavy Lifter AI takes on the complexity. You bring the insight, oversight, and strategic direction. 👉 100% – Full AI Execution In some cases, AI completes the task. But the human still sets the brief, guides the tone, and makes the decisions. For Marketers, Consultants, and Business Owners, this is not the time to resist. It’s time to rethink your value, redistribute your time, and reimagine your workflows. Humans won’t be replaced.... But we will be expected to become smarter — by collaborating with AI. Of course, there’s a much bigger discussion to be had. This post is just one thought in that wider conversation. But it’s a good place to start. With Positivity, Imran. #AIinMarketing #MarketingLeadership #CharteredInstituteOfMarketing #FutureOfWork #AIMarketer

A nice review article "Transforming Science with Large Language Models: A Survey on AI-assisted Scientific Discovery, Experimentation, Content Generation, and Evaluation" covers the scope of tools and approaches for how AI can support science. Some of areas the paper covers: (link in comments) 🔎 Literature search and summarization. Traditional academic search engines rely on keyword-based retrieval, but AI-powered tools such as Elicit and SciSpace enhance search efficiency with semantic analysis, summarization, and citation graph-based recommendations. These tools help researchers sift through vast scientific literature quickly and extract key insights, reducing the time required to identify relevant studies. 💡 Hypothesis generation and idea formation. AI models are being used to analyze scientific literature, extract key themes, and generate novel research hypotheses. Some approaches integrate structured knowledge graphs to ground hypotheses in existing scientific knowledge, reducing the risk of hallucinations. AI-generated hypotheses are evaluated for novelty, relevance, significance, and verifiability, with mixed results depending on domain expertise. 🧪 Scientific experimentation. AI systems are increasingly used to design experiments, execute simulations, and analyze results. Multi-agent frameworks, tree search algorithms, and iterative refinement methods help automate complex workflows. Some AI tools assist in hyperparameter tuning, experiment planning, and even code execution, accelerating the research process. 📊 Data analysis and hypothesis validation. AI-driven tools process vast datasets, identify patterns, and validate hypotheses across disciplines. Benchmarks like SciMON (NLP), TOMATO-Chem (chemistry), and LLM4BioHypoGen (medicine) provide structured datasets for AI-assisted discovery. However, issues like data biases, incomplete records, and privacy concerns remain key challenges. ✍️ Scientific content generation. LLMs help draft papers, generate abstracts, suggest citations, and create scientific figures. Tools like AutomaTikZ convert equations into LaTeX, while AI writing assistants improve clarity. Despite these benefits, risks of AI-generated misinformation, plagiarism, and loss of human creativity raise ethical concerns. 📝 Peer review process. Automated review tools analyze papers, flag inconsistencies, and verify claims. AI-based meta-review generators assist in assessing manuscript quality, potentially reducing bias and improving efficiency. However, AI struggles with nuanced judgment and may reinforce biases in training data. ⚖️ Ethical concerns. AI-assisted scientific workflows pose risks, such as bias in hypothesis generation, lack of transparency in automated experiments, and potential reinforcement of dominant research paradigms while neglecting novel ideas. There are also concerns about the overreliance on AI for critical scientific tasks, potentially compromising research integrity and human oversight.

Google’s Agent2Agent Protocol Is a Big Deal, But It’s Missing One Thing Google’s new Agent2Agent protocol gives agents a shared language to collaborate, discovering each other, negotiating tasks, and working together across frameworks and vendors. It’s a bold step toward fixing the AI silo problem that I've previously written about. Just like HTTP made the web interoperable, A2A aims to do the same for agents. And paired with Anthropic’s MCP (which standardizes tool use), we’re starting to see the foundations of a true multi-agent ecosystem. But there’s a catch. A2A still relies on point-to-point communication (i.e. HTTP, SSE, webhooks). That works at small scale. But in real enterprise environments, it doesn’t hold up. Here’s why: • Too many direct connections (NxM problem) • Tightly coupled agents = brittle systems • Hard to coordinate multi-agent workflows in real time This is the same problem microservices faced and the solution is the same: event-driven architecture. By running A2A communication over Apache Kafka, we get: • Loose coupling between agents • Multiple consumers of every message (agents, logging, analytics) • Durable, replayable events for audit and recovery • Real-time orchestration without custom glue code Kafka doesn’t replace A2A. It fills in what’s missing, making it easier to scale, observe, and connect agents as part of a larger system. Check out my article linked in the comments for full details on this idea.

AI is not just saving us time. It's changing what's possible. Workers using GenAI were 33%  more productive per hour. I think it's even more than that. That is the shift. Not just faster. More output. Better work. More momentum. --- A few years ago, one presentation could take me hours. Now it can take minutes of my time. And it is not just faster. The deck is often more complete. More polished. More thoughtful from the start. --- I can create customer interactive apps in hours. In the past, that would have taken me more than a day or not at all. They are more detailed. More structured. And easier to tailor. --- My favorite tool is Wispr Flow. I do not type. I dictate everything. My grammar and punctuation are corrected. Filler words are removed. Wispr Flow removes friction from almost everything I do. Which means I can capture more ideas, more context, and more depth without slowing down. --- And here is a big win recently. I needed a business workflow connecting Calendly, my CRM, Google Sheets, and Gmail. 1. I asked Gemini to build the workflow instructions to automate the process. It gave me the tools. Then I asked Gemini to use n8n. It generated the prompt to create workflow. 2. I pasted the workflow from Gemini into n8n’s AI and asked it to create it. 3. Within minutes, n8n created the full automation. 4. All I had to do was enter the credentials for each tool and make a few text changes. DONE. ✔️ What normally would have taken me a day took less than 2 hours. And the bigger win was this. I was able to build something I may have put off before. Something more complete. More connected. And more useful. --- Human judgment is still needed. You still have to verify, test, refine, and make sure the output is accurate, useful, and aligned. That is the part more people need to notice. This is not only about moving faster. It is about creating more, finishing more, and freeing up energy for work that matters most. ⚡ USE AI TO DO MORE • Start With One Repeat Task → Pick one thing you do often → Use AI to speed it up and strengthen it • Use AI To Go Deeper → Ask for more detail, structure, and ideas → Not just a faster first draft • Turn Ideas Into Action Faster → Use voice, prompts, and automation → So good ideas do not sit unfinished • Build Better Before You Build Bigger → Let AI help you think through the process → Before you spend hours doing it manually The real win is not speed alone. It is capacity. It is being able to produce more value, with more depth, in less time. If you are ready to save time, increase output, and lead with AI, this is for you. Use, work with, and lead with AI. Join my free AI Readiness Masterclass: 🔗 stephanieshills.com/ai ♻ Repost this for someone still doing work manually 👋 Follow Stephanie Hills, Ph.D. for practical AI, leadership, and career growth tips

It's not sexy to say, but most of AI transformation has nothing to do with AI. There are 10 steps in the sequence of making an internal process or external product AI-native. Only 1 step is AI, and ironically, the other 9 steps are the far harder part. Step 1: Identify the problem - Find the manual process worth automating. turn your brain off autopilot & turn on your "suck meter". - Funny enough, your company becomes more efficient just by mapping out your processes even if you don't introduce AI. Step 2: Understand the workflow - Map how people actually work today. grab an 8.5x11 piece of paper or Excalidraw and create a flow chart of the workflow from beginning to end. - Least sexy part, but generally where the people driving transformation (FDE, GTM engineer, etc) should spend the majority of their time. Step 3: Collect the data - Gather sample inputs, documents, edge cases - Example: for my content machine ai workflow, I gathered past slack messages/notion transcripts to test automated ideation Step 4: Build the prototype [The AI Part] - Whether its engineer-led or SME-led the goal is to test your hypothesis that there's a better way of doing things for yourself as customer zero. Don't worry about code cleanliness, don't worry about scalability. Step 5: Test & iterate - Before you take the process from single player (only you using it) to multiplayer (many users), you want to beat it up with as many rounds of work & feedback + edge cases as possible. Turning every process into a self-improving loop before scaling is key. Step 6: Integrate with systems - Point-in-time data is good for testing the workflow, but live data is necessary before going into production. Step 7: Roll out & train - Whether the new process lives on a live link, on GitHub or an internal library, next step is hand-holding your peers/users through the onboarding process of your new workflow/product. Step 8: Drive adoption - Embed the workflow in your culture where adoption is tracked, ideas & feedback are celebrated, and new/creative use cases become social currency in your business. Step 9: Empower contribution - Treat your new process like an opensource project. Allow users to become contributors. Whether they are literally pushing code or are simply empowered to add ideas/feedback to a kanban board that gets serviced by engineers, make everyone feel like a builder. Step 10: Measure & capture value - If you're in the experimental phase of AI adoption in your company, fuck ROI. The goal is to empower people to throw a lot of shit at the wall & see what's worth focusing on. You don't need to be scientific during this process. - If you're in the scale-up phase of AI in your business, and you need to realize hard ROI, you need to reskill employees attached to this process, undershoot your approved hiring roadmap, or measurably increase ACV/conversion rate/sales cycle speed.