Salesforce tdx2025
https://cloudwerx.co/articles/how-salesforce-agentforce-is-revolutionising-ai-driven-customer-service-what-you-need-to-know/

Salesforce tdx2025

While there were a myriad of demonstrations on new and improved capabilities on the salesforce platform such as the new lightning system design 2 that is deemed to enhance the user experience or? the compression namespace that allows you to interact with zip files directly in apex, the star of the show was agentforce.

I couldn’t resist getting my hands dirty and I ended up completing a mini hackathon that involved using agents to create a flight booking web app that could guide a user through checking for available flights, creating flight reservations and generating documents for that flight booking. Below is a summary of the pros and cons I noticed while completing the development:

  • Pros:
  • Rapid development: By writing instructions in English in the agent builder, I created business logic that would have otherwise been complex and time consuming to create in apex or flows. By doing this you can speed up your development as you don’t need to come up with complex conditional logics in flows or apex, enabling better modularization on your flows or apex to perform only specific actions.
  • Great user experience: I didn’t need to create complex screen flows or lwc components with input fields to get the user input. Instead of all that “noise” that comes with multiple fields on the screen, the end user interacted with just one input field that could take any data and convert it to the right data type for further processing.
  • Using salesforce data to ground your AI responses: While this was not part of the hackathon, it was great to learn through one of the session that you can ground your AI agent to use your case data in salesforce by through clicks and configurations, enabling your service representatives to have better context.


  • Cons:
  • While the use of instructions to orchestrate business logic makes your flows and apex more modular, there is a high risk of having duplicated flows and apex code performing the same actions, especially if you are running an enterprise scale development project with several developers. The good news is that if you are using repositories as your source of truth, you can use AI to perform code validations and check for similarities across your metadata and flag them for resolution.
  • Parallel developments of agents could be complex. Currently, it’s cumbersome and quite a nightmare solving merge conflicts on flow components and I foresee the same problem impacting development on agents.
  • This is not really a con, but I find the notion that developers are no longer needed due to AI to be misguided, at least at this point ??. Using this hackathon exercise as an example, I had to create an integration in apex code to send data and generate documents from DocuSign, configure a flow to embed the agent, configure another flow to query accounts, configure omnichannel settings and several other configurations that I can’t even remember today. My take, if you are a developer, AI will only make you better (if you learn how to use it, if you don’t, well….).

Apart from all the great learnings I had, it was a great experience to connect with my colleagues Wouter Berkelder Ricardo Bajo Ramos Roxana Sperlea Bob Zondervan Bianca Ciuc? as well as other salesforce enthusiasts from all over the world! Looking forward to whats ahead of us! #oneTeam #NN

?

要查看或添加评论,请登录

社区洞察

其他会员也浏览了