Evan Smith’s Post

Check out this great survey of the current landscape with the insightful guidance of a master!

Hours ago engineers led by Srini Tallapragada released a 20K-word manifesto on the quiet transformation of the Salesforce platform over the past 4 years. Here’s what struck me. 85% of customers are already on this Platform – big P as in Platform-as-a-Service (PaaS) – and #Agentforce is the latest manifestation of a march that began in 2008 with Force.com … and before that of course with #metadata and multi-tenancy. Salesforce is not just a salesforce. Salesforce Tech is 300 teams at 23 sites in 14 countries stewarding 200 releases and 250K system changes per week. Environmental changes in recent years include the hyperscalers (Amazon Web Services (AWS), Google, #Azure), regulatory and residency demands, real-time, challenges around resilience and cybersecurity and rampant AI. The 4-year shift is mostly about moving everything to the cloud, breaking monoliths into services, running data lakehouses alongside databases, and of course powering AI. The biggest changes I think are these: 1. #Hyperforce Designed to operate across hyperscalers in 20 regions, its point is to insulate customers from details so they communicate with Salesforce domains. As our co-founder Parker Harris has said, #Hyperforce is “just software” – software that lets the platform run on distributed instances across the globe based on customer requirements. #Hyperforce already uses an AIOps agent to deal with incidents. The agent scans logs, diagnostics, etc., and as of now can detect 82% of CRM incidents and resolve 61% of them automatically, without stressing a human. 2. Data Cloud Data Cloud – as loyal readers know – uses a big data-type #lakehouse, and together with the Salesforce Database comprises the platform. It is built on #Hyperforce (above) and is the foundation for #AI and #analytics. Data Cloud subsumes the CDP and is built w/ Iceberg and Parquet. Cloud-based lakes (AWS, GCP, Azure) use files and folders, and Data Cloud adds various abstractions to make querying and AI easier. There’s support for Zero Copy incl. from unstructured sources, CDP services (like ID resolution) and a JSON-format Data Graph object. This latter includes a Profile graph for things like purchases, browsing history, etc. Data Graph matters because it’s real-time, used by the (doh) real-time layer but also for #GenAI and to trigger actions, e.g., in a #Flow. There’s also the #unstructured data battle, and there are a lot of conversations in the halls about #chunking, #embeddings, keyword indexes, vector indexes, #Milvus, headless semantic layer – which can enrich models with business taxonomies like measures and metrics using a simple declarative language also known as English. #Agentforce will require another post … tomorrow. In the meantime, check out the original white paper at the link in the comments and order my book “Customer 360” from Amazon at another link below – it covers these topics and more!

Recently engineers led by Srini Tallapragada released a 20K-word manifesto on the quiet transformation of the Salesforce platform over the past 4 years. Here’s what struck me. 85% of customers are already on this Platform – big P as in Platform-as-a-Service (PaaS) – and #Agentforce is the latest manifestation of a march that began in 2008 with Force.com … and before that of course with #metadata and multi-tenancy. Salesforce is not just a salesforce. Salesforce Tech is 300 teams at 23 sites in 14 countries stewarding 200 releases and 250K system changes per week. Environmental changes in recent years include the hyperscalers (Amazon Web Services (AWS), Google, #Azure), regulatory and residency demands, real-time, challenges around resilience and cybersecurity and rampant #AI. The 4-year shift is mostly about moving everything to the cloud, breaking monoliths into services, running data #lakehouses alongside databases, and of course powering AI. The biggest changes I think are these: 1. #Hyperforce Designed to operate across hyperscalers in 20 regions, its point is to insulate customers from details so they communicate with Salesforce domains. As our co-founder Parker Harris has said, #Hyperforce is “just software” – software that lets the platform run on distributed instances across the globe based on customer requirements. #Hyperforce already uses an AIOps agent to deal with incidents. The agent scans logs, diagnostics, etc., and as of now can detect 82% of CRM incidents and resolve 61% of them automatically, without stressing a human. 2. Data Cloud Data Cloud – as loyal readers know – uses a big data-type #lakehouse, and together with the Salesforce Database comprises the platform. It is built on #Hyperforce (above) and is the foundation for #AI and #analytics. Data Cloud subsumes the #CDP and is built w/ Iceberg and Parquet. Cloud-based lakes (AWS, GCP, Azure) use files and folders, and Data Cloud adds various abstractions to make querying and AI easier. There’s support for Zero Copy incl. from unstructured sources, CDP services (like ID resolution) and a JSON-format Data Graph object. This latter includes a Profile graph for things like purchases, browsing history, etc. Data Graph matters because it’s real-time, used by the (doh) real-time layer but also for #GenAI and to trigger actions, e.g., in a #Flow. There’s also the #unstructured data battle, and there are a lot of conversations in the halls about #chunking, #embeddings, keyword indexes, vector indexes, #Milvus, headless semantic layer – which can enrich models with business taxonomies like measures and metrics using a simple declarative language also known as English. #Agentforce will require another post … tomorrow. In the meantime, check out the original white paper at the link in the comments and order my book “Customer 360” from Amazon at another link below – it covers these topics and more!

Unlocking data in Datalakes can be a challenging task, often requiring complex datapipes making it hard to use this data in the flow of work. Each movement of data also has the potential to create another data silo. Zero ETL is based on an open source standard that allows customers leverage their Datalake investments by making that data usable in the flow of work without the need to move their data. Salesforce has recently announced their Zero Copy Partner Network which includes AWS, Snowflake, Databricks and now Microsoft. Find out more https://lnkd.in/esjSapSX #Salesforce #DataAnalytics #DataManagement #ZeroCopyPartnerNetwork

🚀 Here comes Salesforce Zero Copy Partner Network! Why should you care ? Short answer is it's a game changer. Some key highlights to share - Seamless Integration: Access live external data securely with zero copy integrations directly into Salesforce Data Cloud. No need for reverse ETL! Enhanced Capabilities: Use data for diverse applications like business intelligence (BI), AI insights, marketing, unified customer experiences, and more, all within Salesforce’s robust metadata framework. Bidirectional Data Flow: Not only fetch but also share crucial insights and unified data back to your systems without duplicating data. Powerful Partnerships: In collaboration with top industry leaders like Accenture, Cognizant, Deloitte Digital, and PwC, and integrating directly with platforms like Snowflake, Google BigQuery, and soon, Databricks and Amazon Redshift. Trusted AI Foundation: Clean and unify data in Data Cloud to ensure AI models are built on the most accurate information, unlocking the full potential of large language models (LLMs) for faster, cost-effective solutions. Expanding Horizons: Support for Zero copy with Heroku Postgres and enhanced integrations through a growing partnership with Microsoft Azure. 🌟 Stay tuned as we revolutionize CRM with Data + AI + Trust. The future is here! #Salesforce #DataCloud #ZeroCopyIntegration #AI

Salesforce made a significant announcement today with the launch of its Zero Copy Partner Network. This new ecosystem aims to enable data sharing and integration across different platforms without requiring data copying or movement. Over the past decade, enterprises have invested heavily in deploying data warehouses/lakes/platforms to consolidate customer information from multiple sources. Initially focused on a few use cases, this centralized data approach worked well... until enterprises found themselves multiplying such repositories and moving massive amounts of data. Businesses are looking for alternate approaches. To make data available across these siloed repositories to multiple applications, some are turning to reverse-ETL (extract, transform, load). With AI, applications and use cases are getting hungrier for more data, making the problem worse. Additionally, AI requires robust data governance, security, and compliance controls - an even greater challenge with replicated data scattered across the enterprise. Zero-copy data approaches help address these data-sharing and integration needs without physically replicating datasets.   Salesforce's announcement of its Zero Copy Partner Network is significant due to the initial partners involved - Amazon Web Services (AWS), Databricks, Google Cloud, Snowflake, and Microsoft. These companies are leaders in technologies for modern data architectures. Furthermore, Salesforce's support of the open-source Apache Iceberg format solidifies Iceberg as an emerging standard. Link to the announcement in the first comment of this post. #moderndataarchitecture #ai #cx

#DataCloud is the backbone of #Salesforce 's push into #AI. It’s a complicated technology, and while many consultants are now Data Cloud certified, there haven’t been many opportunities to put these skills into practice. I've just attended a webinar as part of the #MelbourneSalesforceUserGroup on the key learnings from a three week DataCloud Implementation to harmonize data between Salesforce and #Hubspot. Here’s what I found interesting: Planning the implementation: 💡 Keep it as simple as possible 💡 Work with a defined use case in mind 💡 Consider provisioning times as part of the project timeline 💡 Make sure you have the right people with the right skills to do the implementation. This includes (amount other things) understanding, mapping and transforming the data. Minimizing the credits consumed As Data Cloud operates on usage-based billing, consultants need to find ways to minimise the credits consumed, especially during testing phases. There were a few suggestions to do this such as: 💡 Using a one week trailhead trial account for a dry run (Zero credits are consumed by doing this) 💡 Limiting the scope of your ingestion to minimise the credits consumed by looking at the data you are bringing in and carefully identifying only relevant fields. A data dictionary, or app exchange app can help with this. 💡 Turning off auto-matching in Data Cloud as this is the most expensive process to run in Data Cloud Working with Data Cloud There were also some pointers on implementing data cloud for the first time including: 💡 Using a free app on the app exchange to create your data dictionary. This saves time and also has additional data points such as how often a field is filled in. This can be used to limit the scope of what is ingested, which among other things will save credits.  💡 Make sure you have considered your Salesforce org wide security before implementing, as they will determine what data cloud connector type to use. Organisations that are set to “Private, Private” by default will need to setup a custom connector. 💡 If you are using a field like a phone number or an email address as a primary key, you’ll need to match it to the party field otherwise you won’t be able to use these fields as part of your matching rules 💡 Whenever you update your mappings, you’ll need to refresh them to see the updated mappings. Thank you to PeerNova Inc and Modelit for running the session. I'll post a link to the recording in the comments when it's available. To anyone else out there who has worked on an actual Data Cloud implementation, I'd be keen to hear of any tips, tricks and pointers you have.

A frequently posed question regarding Salesforce's #DataCloud is, "Why not develop this in-house? One can utilize Snowflake, Databricks, or managed SQL to create integrations or incorporate other off-the-shelf tools as needed." Before addressing this, let us consider a seemingly unrelated topic. Recently, #StackOverflow published the findings of their 2024 Developer Survey, which highlighted the "most common frustrations" among developers. Notably, 62% of respondents identified "Technical debt" as their primary concern, with the complexity of technology stacks for building and deployment ranking second and third. In terms of required data sources, Data Cloud does not contribute to the technical debt associated with custom integrations or proprietary coding, particularly for organisations already embedded in the Salesforce ecosystem. Almost all Data Cloud projects I have been a art of utilize out-of-the-box connectors, drag-and-drop mapping and transformation, and declarative solutions for data activation. This methodology is thoroughly documented and supported by Salesforce. In my view, among the numerous advantages, this represents a highly compelling rationale for organisations to evaluate Data Cloud as a robust, no-code, and scalable solution for unifying and optimising their data assets. Developing a custom solution may be justifiable if an organization possesses the requisite in-house expertise and budgetary resources. However, if such capabilities are lacking, it is prudent to avoid incurring additional personnel overhead when a viable solution is readily available. Data Cloud serves as a seamless extension of the Salesforce platform with which organisations are already acquainted, thereby facilitating efficient data management without the complexities associated with bespoke development. #Salesforce #Trailblazer #Trailhead #DataCloud #AI #GoldenRecord #ZeroCopy #Analytics #ZeroCode https://lnkd.in/gpWurzmy

Tired of data import woes? Skip Sauls shares how Data Cloud’s AI simplifies external data integration for Salesforce #AwesomeAdmins. No more back-end headaches! Learn how:

Tired of data import woes? Skip Sauls shares how Data Cloud’s AI simplifies external data integration for Salesforce #AwesomeAdmins. No more back-end headaches! Learn how: