In the latest iDialogue Flow CPQ+AI release, transactions are divided into guided chat prompts and Auto-Launched flows. This is the traditional Model-View-Controller (MVC) development pattern many developers are familiar with, where the “View” is implemented as a GPT AI Agent and the Controller is an auto-launched flow that handles validation and record INSERT. This division of responsibility aligns well with the job functions of Business Analyst (BA) and Salesforce Developer; where the BA defines product bundles and add-ons using natural language expressions that are passed to Flows for more deterministic validation and record handling. This implementation is far faster because the BA's requirements and specifications are immediately executable "code" that offers real-time feedback.
Michael Leach’s Post
More Relevant Posts
-
When you have huge volumes of 100% accurate data AI can do amazing things. Org metadata is 100% accurate, because it is the config of your org. At Texas Dreamin' I presented a number of real use cases for AI for business analysis and metadata management. This is such an exciting time. Prompts that can be used in ElemerntsGPT and ChatGPT - field complexity analysis - metadata description quality evaluation - field help text quality evaluation - auto-generation of process maps - creation of user stories - evaluation of process diagrams, flowcharts, BMN & UML diagrams - getting process diagram insights We'll be launching these on https://lnkd.in/gV53y8f9 over the coming days. Field complexity analysis is being published today. Subscribe to get them. #AI #metadata #salesforce #prompt
To view or add a comment, sign in
-
-
Craving more automation? See first hand how Mulesoft Intelligent Document Processing (IDP) can quickly extract and organize unstructured data from various document formats including PDFs and images using pre-built templates and natural language prompts powered by Einstein. #automation #mulesoft #IDP #AI #documentprocessing
To view or add a comment, sign in
-
Salesforce Prompt Builder seems complex 😟 The Simplicity vs. Complexity Debate As I watch tutorials on using Salesforce Prompt Builder, I'm left wondering: do we really need to go through all the parameters to create a simple prompt? With advanced models like Llama3, GPT4, or Claude 3, can't we just provide a description of our offers and let the AI generate an informed response? This looks like a waste of time and a lot of consulting budgets will be spent on something that current AI can do. Salesforce engineers are underestimating what a model can do. The current models can choose which fields to pick to generate content. They might even be better than the engineers in picking the right fields. Give an agent access to the database and it will figure it out by itself. Let us make things simpler. It seems to me that there are two approaches to AI development: the engineer approach, where we design the entire process, and the agent approach, where we simply instruct the AI on what to do. I'd love to hear from others: shouldn't AI be making our lives easier, not adding to the complexity we already face? Share your thoughts on the ideal AI development approach and how we can strike a balance between simplicity and effectiveness. (c) crafted with Llama3 70B #Salesforce #PromptBuilder #AIDevelopment #SimplicityVsComplexity #ConversationAI #LLaMA3 #GPT4 #Claude3
To view or add a comment, sign in
-
-
👉 "What it means for us as Developers and Admins is that it's going to change what we build and also how we build it." -Clara Shih during #TDX24 The slide below is a simple graphic, but study it for a second. In addition to how we build, it represents a fundamental step change in how AI is reshaping how we engage with the #Salesforce platform. • Navigation → Conversation We currently navigate Salesforce to: locate information, transition screen to screen, make settings changes, update data, and so much more. We are transitioning to a conversational user interface where you'll perform actions across the platform with natural language speech or text. • Hard-coded → Dynamic Currently, code is logical. It's set and static. Decision paths are predefined and expected to execute when conditions are met. If this then that. We are transitioning to business logic building blocks that include AI which has reasoning capability and is dynamic. Conditions may or may not have been met. AI will tell us! It will advise on the "next best action(s)" 😜 for the typical coded workflow. • Manual → Augmented With new AI capabilities, we have a grand-canyon-sized opportunity to augment human activities across the platform: evaluation, decision-making, actions, reactions, workflows, auditing, scheduling, notifications, follow-ups, case work, sales calls, service calls, messaging, configuration changes, and the list goes on and on. 🌟 Follow me for more! 🌟 Don't forget to to complete the AI Skills Quest on #Trailhead by the end of April for a chance to win a free voucher for your AI Associate certification! 🎉
To view or add a comment, sign in
-
-
Health at Every Size ☮️ Marketers with SaaS Co-Founder 💋 | Chief Content Officer at diem 🌄 AI Researcher & Coach | IG 📷 @brokegirlbranding | Cold Brews | Poetry Educator
Recap: Salesforce World Tour NY Highlights 🌟 👀 Favorite Sessions: Tied for first place! AI Hallucinations with Itai Asseo, a leader in AI at Salesforce, and Buyers Edge Platform use cases for Data Cloud and MuleSoft with Sean Donahue. 🔍 Top Picks: Session on RAG (Retrieval-Augmented Generation): Ensemble RAG: A blend of retrieval techniques ensuring the most reliable data surfaces. Techniques Explained: 1)Embedding Vectors: Utilizes enhanced metadata for accuracy. 2)Knowledge Graphs: Improves interpretability with structured data (Subject, Predicate, Object). 3)Recursive Hierarchical Trees: Aids in contextual retrieval across extensive documents (ideal for legal documents and others). 💡 Insights Gained: Quality Assurance: Suggest integrating subject matter experts to refine AI models and enhance semantic understanding. User Interface Considerations: Usage Context: Would these comparisons work better on larger screens? What would the UI need to look like for mobile? Design Simplicity: Excessive motion can be distracting. Example: ChatGPT 4 comparison on desktop was hard to follow due to simultaneous content presentation and different generation speeds. Hard to process text as it's being generated or to hone in a single version at a time. Accessibility: Visual Design: Experiment with font weights and colors to enhance readability across various modes, adhering to ADA and WebAIM guidelines. 🏷️ Tags: #AI #Salesforce #DataCloud 🔄 Engage & Share: What are your thoughts on the future of AI in user interfaces? Join the discussion below! ⬇️
Health at Every Size ☮️ Marketers with SaaS Co-Founder 💋 | Chief Content Officer at diem 🌄 AI Researcher & Coach | IG 📷 @brokegirlbranding | Cold Brews | Poetry Educator
That's a wrap. I went to a lot of sessions at Salesforce World Tour NY. There was a great dive into RAG (Retrieval-Augmented Generation) with Itai Asseo, an AI Leader at Salesforce. Doing some dope things in AI Research. In short, he spoke about Ensemble RAG, which blends retrieval techniques. The three methods are as follows: 1. Embedding Vectors - Embedding vectors with enhanced metadata 2. Knowledge Graphs - Enhances interpretability with knowledge triplets (Subject, Predicate Object) 3. Recursive Hierarchical Tress - Empowers contextual retrievals across lengthy documents (I'm looking at you, Law) This threesome approach runs and the winner is what gets presented to the user. So, it takes some of the guessing game out of the... "is this *#&! for real? Can I trust this?" My brain droppings: 🧠 💩 1. Have subject matter experts revise the models. Improve the semantic associations/clusters/hierarchies (I don't know the technical term). Create super individualized models for particular types of processes/content types. 2. UI... where will people use this mobile? Desktop? Is too much movement a bad thing? Can it distract the user? Example: ChatGPT 4 on the desktop asked me to rate which version I preferred. The content was generated side by side at different speeds. SUPER distracting. Ultra hard to process. And, hard to focus on a singular's prompts content in this design format. 3) Dark mode??? We shall see. It would be interesting to experiment more with font weights and colors with AI prompt responses. While being mindful of ADA and WebAIM best practices across light and dark toggles. #AI #Salesforce #DataCloud
SALESFORCE WORLD TOUR BREAKDOWN
www.linkedin.com
To view or add a comment, sign in
-
I'm excited to share a new video about Visual Question Answering (VQA) using the Salesforce model from Hugging Face! Visual Question Answering (VQA) is a fascinating AI field where machines answer questions based on images. I've used the powerful Salesforce model to create a system that understands and responds to visual content. This technology can transform various industries, from e-commerce to healthcare. The potential applications are endless and incredibly impactful. Can't wait to hear your thoughts! Check out the video to see how it works! AIMER Society - Artificial Intelligence Medical and Engineering Researchers Society #Sai_Satish sir #Aimer #VisualQuestionAnswering #AIResearch #MachineLearning #ComputerVision #NaturalLanguageProcessing #HuggingFace #Research #ArtificialIntelligence
To view or add a comment, sign in
-
Use AI to Create Solutions for Technical Support Cases! Watch Metadata Studio automatically create Solutions by analyzing technical support Cases! If the Case was resolved, a Solution is created, otherwise the Cases are sorted into categories for your product, engineering, website, or sales team. This is next generation technology for org documentation, support automation, and technical debt reduction! Happening fast on OpenAI by Metazoa for Salesforce... #promptengineering #chatgpt #artificialintelligence #ai #salesforce #salesforceadmin #customersuccess #salesforcedeveloper #Metadatastudio #snapshotai
To view or add a comment, sign in
-
Salesforce Email-to-Case is the perfect candidate for transformation leveraging Generative AI. It will save 💰 + Significantly improve the turnaround time ⏳ Using GPTfy(Salesforce AppExchange) auto-populate the routing criteria: - Product - Category (Warranty Inquiry, Software Bug, Hardware Failure, Billing) - Priority - Language - Region, etc. Once populated, your existing match-making logic will take over and assign the case. All of this will happen in near-real-time 🚀
To view or add a comment, sign in
-
-
🔥 Advanced Techniques for Training Einstein Bots 🔥 💡 Hey Salesforce Developers and Admins! Want to enhance your Einstein Bots' conversational capabilities and level up your customer interactions? 🤝 Look no further! Check out these advanced techniques to optimize your training process: 🎯 **1. Intent Set Creation and Management** - Organize Intent Sets: Keep your intents neatly structured for easier management and updates. 🏢 - Use Intent Hierarchy: Group similar intents under parent intents to improve context understanding. 🌳 🗂️ **2. Dialog Design and Configuration** - Use Conditional Logic: Handle complex scenarios and provide precise responses using variables, loops, and conditional statements. 🔄 - Integrate with Salesforce Data: Personalize conversations by leveraging Salesforce data, including custom objects and fields. 📊 🚀 **3. Bot Training and Testing** - Use Real-World Data: Train your bot using historical interactions or mock user inputs for accurate responses. 📚 - Test and Refine: Continuously analyze metrics like intent recognition accuracy and user satisfaction to optimize performance. 🧪 🔤 **4. Advanced NLP Techniques** - Entity Recognition: Implement entity recognition to identify specific entities within user inputs, improving accuracy. 🕵️ - Contextual Understanding: Analyze conversation history and adapt responses accordingly for personalized interactions. 🗣️ 💼 **5. Integration with Other Salesforce Features** - Service Cloud Integration: Seamlessly automate common tasks and enhance customer service. 🌐 - Salesforce Flows: Automate complex business processes and integrate them with your Einstein Bot. 🔄 🔗 By applying these advanced techniques, you can elevate the performance and effectiveness of your Einstein Bots. 👌 Enjoy enhanced customer experiences and increased efficiency in your contact center operations! 🛠️ #SalesforceDeveloper #SalesforceAdmin #EinsteinBots #AdvancedTechniques #CustomerExperience #Efficiency #Optimization #Automation
To view or add a comment, sign in
-
Accelerate Salesforce releases with Functionize AI testing powered by advanced deep learning models and massive cloud-scale data. - Rapid Builds: Quickly develop a comprehensive automation suite to validate customizations and third-party integrations. - Role-Based Testing: Ensure consistent test data and user permissions aligned with your business needs, with native test data generation and TDM. - Transition to Lightning: Ease your move to Salesforce Lightning with automated tests created in minutes. Learn more about how to modernize your Salesforce testing: https://lnkd.in/ek_D3km8 #Salesforce #SoftwareQuality #SoftwareTesting #TestAutomation #AutomatedTesting #SoftwareQA #QATesting #DeepLearning #GenAI
Salesforce Testing | Functionize
To view or add a comment, sign in
Salesforce Engineering Director | Business Application Development | Working with AI | chatCPt
3moMichael Leach can you share more details around which industries are adopting this very forward thinking quoting and pricing architecture? Would love to hear more about where the pricing data is stored (presumably in a standard price book object) as well.