AI Moderation for Large-Scale Panels: Human Moderators Still the Best Choice

A lot of practices I speak to aren’t short of work or ideas. The issue is that too much knowledge is stuck in people’s heads, too many processes rely on repetition, and too much senior time is spent on work that could be structured better. That’s where AI can become useful. Not as a gimmick, but as a way to reduce friction in the practice. The question isn’t really “should we be using AI?” It’s “where could AI help us remove friction without adding complexity?” That’s the conversation more practices need to start having.

AI workflows are becoming more sophisticated with the introduction of persistent context. This persistent context can be fed into every workflow, allowing for a step-by-step, prompt-by-prompt execution. For example, if you define your entire workflow as persistent context and then ask the AI to run it on specific details, it will process through each stage to achieve the desired outcome. Look for features that allow you to attach files and enter prompts, then save them—these are likely the 'projects,' 'artifacts,' or 'gems' that enable these advanced AI capabilities. #AI #ArtificialIntelligence #WorkflowAutomation #MachineLearning #TechInnovation

You changed the prompt. You switched the model. But how do you know your AI system actually improved? That's where evaluation comes in. While building my Weekly Goal Planner Agent, I came across a simple AI evaluation framework that every production AI system should have. In this carousel, I share a simple framework for evaluating LLM applications beyond just accuracy. Project: https://lnkd.in/dv4ymikU What evaluation techniques are you using in your AI projects? #AI #LLM #AIEvaluation #AIEngineering #GenerativeAI #MachineLearning #SoftwareEngineering

AI Mutualism: The notion that we can mindfully use AI to improve ourselves. When working with tools ask: * Please provide constructive feedback on this session * Where can I be more clear? / Ask me clarifying questions * What am I not seeing? etc. If you have any favorites, please comment.

What is the lowest-risk way for a small business to begin using AI? Start with operational clarity. Before buying a tool, identify: 1. Where work gets stuck. 2. Where time is being lost. 3. Where customers or employees feel friction. 4. Where the process is repeatable. 5. Where human judgment must remain central. That gives AI a job to do. Without that clarity, AI becomes another shiny system layered on top of existing confusion. We wrote about lowering the barrier to AI — and why readiness should come before tooling. Read the blog: https://lnkd.in/gBk53w4C Start with the assessment: https://lnkd.in/gn62mFaJ #AIReadiness #SmallBusinessAI #HumanCenteredAI

Without an AI strategy, most investigators are already using consumer AI tools. But in attribution-sensitive investigations, even the earliest queries can reveal research intent. ⚠️ In this clip, we look at where consumer AI breaks down in a typical investigative workflow and what teams need to consider before sensitive information leaves their control. Watch the full on-demand webinar. Link in the comments. #AI #DigitalInvestigation

Clear messaging for the human gaze is one thing, but is it clear to machines too? As people increasingly use GenAI tools to source information, not knowing what information is being found about you when people search has become a blind spot. If messaging is unclear or has gaps, GenAI tools will fill them with incomplete, outdated or inaccurate information. From a reputation perspective, that’s a risk worth paying attention to. Jonathan Cookson recently published a perspective piece, ‘What is GenAI saying about your organisation’, exploring why it’s important to understand how you’re being represented by AI systems. The question is no longer only whether people are searching for your organisation online, it’s whether they’re asking AI about it first. You can read it here https://lnkd.in/gA9ZZGgc

At this month’s Builders’ Breakfast, Nevenka Toma-Vitasovich shared practical insights on the growing role of AI in business. As AI becomes more widely adopted across every industry, it’s important to recognise not only its potential but also its limitations. One of the biggest risks is that AI often sounds confident and convincing - even when it’s wrong. In this presentation, Nevenka explores the question: “Ask not what AI can do for you… but what it can accurately do for you.” It’s a timely reminder that getting the best from AI starts with understanding where it adds value - and where human judgment remains essential. Watch the presentation here: https://lnkd.in/eg4mS2sb #AIinbusiness #AIinconstruction #buildersbreakfast #MartelliMcKegg

AI terminology, simplified. Models, agents, workflows, and agentic applications are often used interchangeably, but they each play a unique role. Establishing clear definitions across organizations helps avoid confusion, so you can apply AI in a meaningful way. Our AI Terminology Guide breaks down essential AI terms, helping organizations build a shared foundation and better understanding of how AI fits into real business processes. Check out the guide here: https://lnkd.in/gbWSyG9s #AI #AITerminology #EnterpriseTechnology #Innovation #TeamElire

Here's a question for procurement professionals... If an evaluator uses AI to draft their evaluation responses, should that be acceptable? The score may still be their own, but if the rationale and narrative have been generated by AI, is that still the evaluator's assessment? Where do we draw the line between using AI as a productivity tool and relying on it to produce part of the evaluation? With AI becoming part of everyday working life, I don't think procurement can avoid having this conversation for much longer. What's your view? 🤔