By Shraddha Nakra and Marlene Guerra-Reeve · Salesforce ArchitectsYou will also see the steps to inventory existing connections and plan a controlled cutover. The piece is aimed at architects and admins who inherit orgs that share data with partners or other orgs.
You will know how Salesforce to Salesforce works, what its retirement means for cross-org sharing, and how to choose among Experience Cloud, Partner Connect, Salesforce Connect, Data 360 and MuleSoft as replacements.
Why read this
You will also see the steps to inventory existing connections and plan a controlled cutover. The piece is aimed at architects and admins who inherit orgs that share data with partners or other orgs.
Inventory every connection in each org
Retire connections no longer needed
Gather business requirements for the rest
Choose a partner or platform option
Plan a controlled implementation
The first step in replacing S2S is inventorying every connection in each org
Salesforce to Salesforce (S2S) is a declarative, point-to-point integration that shares records between two Salesforce orgs without code or middleware. The publishing org chooses what to share, the subscribing org maps it to its own fields, and forwarded records are sent asynchronously with retries and errors logged in the Connection History. Authentication relies on a mutual invitation, so no credentials are managed. Salesforce is retiring S2S with the Spring ’27 release, so teams need a new strategy, and the article suggests using the retirement as a chance to close gaps the current setup leaves open.
The first step is an inventory of every connection in each org. Record what is shared, how records are forwarded and accepted, and who owns each connection, along with the automation, assignment rules and logic that depend on it. Retire connections no longer needed, then gather business requirements for the rest. For partner scenarios, Experience Cloud is the choice when partners can work in your org on records you own, with Default External Access set to Private and access granted through partner roles and Sharing Rules. Partner Cloud adds onboarding, tiers and channel revenue management on top of that site, at the cost of licenses and program design. Partner Connect (Beta) suits partners that need their own synced copy, but it supports only accounts, contacts, leads, opportunities and cases as of Winter ’27, and not formula, multiselect or most lookup fields. MuleSoft can replicate records as an alternative for production.
When no partner-facing option fits, the article works through platform options in order of how much data must move. Salesforce Connect’s cross-org adapter reads or writes records in place as external objects but suits only small, on-demand operations. Data 360 offers a shared view through Data Cloud One or Data 360 to Data 360 sharing, though neither creates or updates CRM records. MuleSoft for Flow or Anypoint Platform handles replicated records where each org keeps its own copy. A custom build, using Apex, REST, Change Data Capture or Bulk API 2.0, is a last resort and takes on duplicate handling, backfill, deletes and credential management.
Finally, the article recommends a controlled implementation. Agree timelines and field ownership with the other org’s owner. Identify every flow, trigger, report and integration that touches the moved data, since automation also runs on records the new integration writes. Test duplicate prevention and failure cases in a sandbox with pass criteria set in advance, and finish production cutover several weeks before the instance’s Spring ’27 upgrade date, without running S2S and the replacement side by side.
You will learn how Carter is structured, with its four preconfigured subagents, page awareness and merchant controls over guided discovery, and what the Shopper Agent template gives you to start from.
Why read this
You will also see the customer results Salesforce reports, though they are vendor claims.
Shopper asks in natural language
Agentic Commerce Search reads the intent
Carter routes to a subagent
Answer grounded in the product catalog
How Carter handles a shopper request across its subagents
Salesforce introduces Carter, a shopper agent from Commerce Cloud that gives online shoppers a conversational interface in place of keyword search and category pages. The article argues that shoppers now expect the guided, context-aware help a good in-store sales associate gives, and that Carter is meant to provide it. It is available through B2C Commerce and can be reached from the search bar, the header icon, or FAQ widgets on product detail pages.
Carter is powered by Agentic Commerce Search, which interprets the intent behind a natural language request rather than matching keywords. Its answers are grounded in the merchant’s product catalog, and brands can add other knowledge sources through Data 360. Merchants control how Carter sells by choosing which follow-up questions are required or optional in guided discovery and which selling points it highlights when comparing products.
Getting started means using the Shopper Agent for B2C Commerce template in Agentforce, which comes with all the subagents and actions preconfigured. Product Expert handles browsing, search, recommendations, comparisons and product questions. Site FAQ covers policies such as shipping, returns and refunds. Cart manages products, variants, promotions and coupons. Order Confirmation confirms orders and recovers carts after failed or cancelled payments. Carter orchestrates these behind one chat, and because it is page-aware it can answer questions about the product or cart the shopper is currently viewing.
The article cites several retailers using Carter under their own brand names, including Cuisinart’s Compass, Hibbett AI and SharkNinja’s combined commerce and support interface. Hibbett reports going live in six weeks and handling 90% of core shopper journeys. These are vendor-supplied results. Carter is generally available now as an add-on for B2C Commerce customers and works in every language Agentforce supports.
You will learn how the team generated synthetic incident data, validated models using scenario tests instead of accuracy alone, and shipped a reusable Data 360 training and scoring pipeline that trains per customer.
Sync incidents and milestones into DMOs
Daily transform labels milestone violations 0/1
Einstein on Data 360 trains XGBoost classifier
Scheduled batch writes scores to score DMO
AI Accelerator card shows score on incident
How the template turns customer incident history into scored predictions on the incident record
Salesforce’s engineering team set out to predict SLA breaches and major incidents in IT service management without customers needing data scientists, and without any real customer incident history to learn from at the start. They built three Predictive Model Templates: SLA Breach Prediction, Major Incident Prediction, and Change Risk Score. Scores and leading predictors are shown on the incident record through Data 360.
Synthetic data came from product manager samples, expanded by varying features and generated with open-source LLMs within product guardrails. The team noted that the simulated Customer Zero data was insufficient on its own, so grounding in realistic scenarios mattered more than volume.
Because SLA breaches are rare, aggregate accuracy can mislead; a naive model could look 95% accurate and still be useless. Instead, the team held all incident features fixed and varied one factor at a time, such as owners with a higher or lower historical milestone miss rate, checking that the score moved in the expected direction. They also checked whether the top three predictors matched what an experienced service manager would flag. The article stresses that these results validate tested scenarios, not production accuracy.
The template ships the training contract rather than fixed weights: DMO API names, transformation logic and model structure. Each customer’s history determines the learned parameters. Incidents, milestones and related records sync into DMOs, a daily transform labels milestone violations as a 0/1 target, and Einstein on Data 360 trains an XGBoost binary classifier. A scoring transform writes predictions to a score DMO, displayed on the AI Accelerator card.
Two constraints remain. Scoring runs as a scheduled batch about once an hour rather than in real time, chosen for scalability with large customers. Training requires roughly 400 completed incidents, and prediction quality depends on the customer’s history. The templates reached general availability in February.
You will learn what the Salesforce and Microsoft integrations now offer: Salesforce context inside Copilot and Teams, and bidirectional zero-copy data sharing between Data 360 and Fabric.
Why read this
You will also see which parts are in beta, pilot or generally available.
Data Federation
Query Fabric OneLake data in Data 360
Source data stays in Fabric
Generally available
Data Sharing
Share curated Data 360 context to OneLake
Supports custom analytics and AI workloads
Public preview, beta
Data Federation lets Salesforce query Fabric data in place, while Data Sharing sends Data 360 context to OneLake
Salesforce and Microsoft have announced Copilotforce, a partnership that brings Salesforce customer context and sales workflows into Microsoft Copilot, adds real-time CRM collaboration in Microsoft Teams, and connects Salesforce Data 360 with Microsoft Fabric. Accenture is the design and implementation partner and is deploying the integrations in its own global sales organization.
Salesforce in Microsoft Copilot lets sellers ask natural-language questions and get grounded outputs, prepare for meetings, analyze opportunities and take supported CRM actions. Copilot’s Work IQ layer combines Salesforce data with Teams meetings, Outlook emails and shared documents on SharePoint and OneDrive. The article names three daily workflows: meeting preparation, deal-health analysis and pipeline review with close plans in Word and PowerPoint.
The Sales Cloud in Teams app, which was shown at Dreamforce, is in pilot. It shows pipeline and quota progress, sends deal notifications, and lets sellers update supported opportunity details through adaptive cards, with changes saved back to Salesforce. Agentforce Coworker handles actions within Teams.
On the data side, Data Federation is generally available and lets joint customers query Microsoft Fabric OneLake data inside Data 360 without moving it. Data Sharing is in public preview and beta and lets organizations share curated Data 360 context into OneLake. Access follows existing Salesforce permissions and Sharing Rules, and supported writeback actions require explicit user confirmation and respect Validation Rules. The article stresses that pricing and availability depend on customer agreements.
You will learn what Salesforce Catalog indexes, how it grounds AI answers and code generation in your org's real object and field names, and how an admin enables it and connects an AI assistant or the Data 360 MCP server.
Without Catalog
Agent guesses field names
Deploy fails on unused field
Engineer or Setup review needed
With Catalog
Agent searches the live index
Returns closest real matches
Answers only what user may see
Catalog lets an agent check the org's real names before it answers or writes code
Salesforce Catalog is an index of how an org is built, covering Objects, Fields, Flows, Apex, Permission Sets, users, and Data 360 components such as data streams, calculated insights and segments. It updates as teams deploy changes, so no separate documentation has to be written or maintained. An AI assistant or agent consults the Catalog before answering, and every question runs as the signed-in user, so answers are limited to what that user may see.
The article gives examples from a service operations manager tracing a CSAT figure back to its source, a data owner checking what depends on a field before changing it, and a developer who avoids using an unused field name. Catalog matches questions by meaning as well as by name, returns a short ranked list of candidates, and lets the agent read details only for those, which the article says keeps responses fast in large orgs and reduces failed deploys.
An admin enables Catalog in Data Cloud Setup and turns on the MCP Server, then connects an AI assistant such as Claude, ChatGPT or Cursor, or Agentforce Vibes, which comes preconfigured. Teams building their own agents can use the Data 360 MCP Server, and Agent Context Engine uses Catalog to supply context. The article notes that Catalog is available in every edition that supports Data 360 and points to a setup guide for steps and permissions.
You will learn what Agentforce Coworker now does across Microsoft Teams, mobile and desktop, what is available now versus later, and how it respects Salesforce permissions and governance.
Employee delegates a job
Coworker plans the steps
Agents and flows run
Results return for review
How Coworker handles a delegated job, as described in the article
Salesforce reports that more than 185,000 users are working with Agentforce Coworker, with usage up 21x in three months. The release adds Conversation History, reusable Skills and scheduled tasks, plus connections to Google Workspace and Microsoft 365 so that email, calendar and file data can sit alongside Salesforce records. Coworker can be used from Microsoft Teams and new iOS, Mac and Windows apps.
Coworker plans multi-step jobs, coordinates Agentforce agents, flows and business actions, and returns results for review. It is described as working within existing Sharing Rules, permissions and configured access controls, with approval and audit capabilities for supported operations. A new $50 per-user, per-month option is offered for predictable budgeting.
Availability is staged: Google Workspace and Microsoft 365 connectivity, Conversation History, Skills and the per-user SKU are available now; the Teams, iOS and desktop apps and a Coworker MCP server are slated for November 2026; Daily Digest, scheduled tasks and mobile apps for Field Service, Consumer Goods Cloud and Life Sciences Cloud are expected in December 2026. The article notes pricing and features may change, and much of the piece is customer testimony and marketing framing.
You will learn how the Metadata Grounding MCP Server indexes org metadata, how to connect it, and how to use describe, search and query so agents reuse existing components instead of rebuilding them.
describe_metadata: list types and schemas
search_metadata: find items by intent
query_metadata: read the full record
The three Metadata Grounding tools run in order, so the agent checks the org before writing code
Coding agents working on a Salesforce org often build new components because they only see the part of the org that has been retrieved into a local project. The Metadata Grounding MCP Server (Beta) addresses this by searching a pre-indexed catalog of the whole org’s metadata, so an agent can check what already exists before writing code. It is one of three pieces in the Headless Toolkit, alongside Skills, which teach patterns, and the Headless 360 MCP Server, which lets the agent act. Metadata Grounding answers the question of what is currently in the org. It is read-only, runs as the authenticated user so it returns only what that user may access, and covers more than 70 core metadata types, though not Data 360 metadata.
The server exposes three tools: describe_metadata lists available types and their schemas, search_metadata finds components by natural-language intent with optional type and subtype filters, and query_metadata returns the full record for a result. The recommended workflow is describe, then search, then query. Behind it, Salesforce Catalog extracts metadata from the org, stores it in a Data 360 data lake object mapped to a Catalog data model, and indexes it with a hybrid of vector and keyword search. Catalog updates supported types incrementally and runs a full refresh every 24 hours, so very recent deployments may take time to appear.
The article illustrates two uses in Pronto, a sample food delivery org. In the first, a vague bug report about a non-working menu category was traced to an Apex controller and a Lightning Web Component found by describing the symptom, with the type filter narrowing noisy results. In the second, an agent found an existing Apex class with a summarize-reviews action and extended it rather than building a duplicate. The article advises adding a project rule telling agents to search the org first. It also covers setup: an org with a supported Data 360 edition, Salesforce Catalog enabled, and admin activation of the server, with the server available either standalone for any MCP client or inside the Headless 360 MCP Server, which will later be renamed the Salesforce MCP Server. Caveats include that search only finds candidates, which should be confirmed by query; the server does not return source code; and usage is metered in production but not in pre-production orgs. Planned additions are AI-generated metadata descriptions and a Metadata Graph for dependency analysis.
You will learn how Salesforce assesses knowledge content before indexing, how it tests Data 360 retrieval changes against a golden data set, and why a human-reviewed failure rate matters more than raw metrics.
Hypothesize how the change affects answers
Build the recipe in Data 360
Evaluate with quantitative and qualitative tests
Reject or deploy, then keep monitoring
The loop used to test each change to the agent's Data 360 configuration
The Salesforce Help Agent is grounded in Salesforce’s own product knowledge, prepared and organized in Data 360 and used by Agentforce through retrieval augmented generation. The authors argue that an agent’s answers can be no better than the content it retrieves, so content quality has to be checked before any model or configuration work starts.
The first lesson is a Fit, Gap, Clash review of every article: whether it answers a real question, whether it leaves a customer issue uncovered, and whether it contradicts other content. Articles are scored using health, failure, resolution and traffic signals, paired with top case drivers. More content does not mean better coverage, and the team chose a separate agent for the Informatica help portal rather than adding that content to the existing Help Agent, which now resolves 92% of questions on that site without a case being filed. Content authors and an AI and Data team share the work, with Agent Managers connecting them.
The second and third lessons describe building by hypothesis: each change to a Data 360 recipe is tested, and only changes that improve answers ship. A product-name metadata hypothesis showed flat results and was rejected, leading to a new recipe that pairs metadata with prompt instructions. Evaluation uses a golden data set of about 200 utterances, each validated by a person, with acceptance criteria covering correct, complete and relevant answers.
The final lesson combines deterministic retrieval metrics, semantic quality checks and business outcomes with human and LLM review. When section-aware chunking was tested against fixed-size chunking on 195 query-document pairs, metrics flagged 8 failures, but review showed the real failure rate was closer to 1%, and that figure was used to approve production.
You will learn how Salesforce built lead scoring and three-way routing on a unified Data 360 profile, and which design choices carried the gains.
Why read this
The piece also gives a practical checklist for applying the pattern at a smaller scale.
Match to unified profile
Classify job level and score
Apply team thresholds
Route to sales, nurture or agent
Each inbound response is matched, scored and sent to one of three routes
Salesforce’s sales team routed leads across three systems with heavily customized rules, and routing changes often went live weeks late. The team unified signals such as website activity, event attendance, email engagement, support history and product usage in Data 360, then built scoring and routing on that profile. Its in-house data science team built a lead score from 1 to 5 and a product interest score, and each inbound response is matched to a profile, classified by job level, scored, and sent to a sales team, marketing nurture or agent-first outreach.
The article reports that lead assignment fell from 20 minutes to 20 seconds, custom code dropped 90% from about 2 million lines, and routing changes now take days instead of four weeks. Thresholds sit in configurable tables rather than code, so a business user could raise the bar for Tableau leads without a release. The authors credit the whole implementation for these results, not Data 360 alone, and note that the same profile drives an email journey with a 233% higher click-through rate than their average Marketing Cloud journey.
The closing advice is to unify data before scoring, define what sales-ready means for each team, give every lead a destination, and keep rule changes out of code releases.
You will learn how weak business context makes Salesforce agents burn far more tokens, and which low-cost habits (documenting objects, defining terms, cleaning records, using Flow where no reasoning is needed) reduce that cost before Headless Platform Interaction metering starts.
43,000: Tokens used without the harness, with a wrong answer
7,000: Tokens used with the configured harness
The same question used far fewer tokens once the agent had proper business context
A Dreamforce demo followed a sales rep asking an agent which customers to focus on to close a Q3 gap. Without a harness, the agent used 43,000 tokens and gave a wrong answer: it recommended a frozen customer, missed a renewal already in the pipeline, and misread fiscal Q3. With a properly configured harness that pulled live pipeline data, used a semantic layer to define Q3, and drew on memories of past customer interactions, the same question took 7,000 tokens and produced a sound answer.
The article argues that Flex Credit burn comes largely from the context an agent must work out for itself. Outcome-based pricing and the planned charge for Headless Platform Interactions make this matter, since the meter runs whether or not the answer is good. No multiplier has been published yet, and Salesforce has promised 30 days’ notice before metering starts. Context packs let a harness supply only the subset of information an agent needs, so the aim is the right context rather than the most context.
The author suggests practical steps for now: fully document the one or two objects agents will rely on most, write down ambiguous definitions such as fiscal quarters, clear stale and duplicate records, use a plain Flow where a deterministic answer is enough, and record a consumption baseline before metering gets more complicated.
You will know how to structure a Tableau Semantics model for query speed, correct joins, governance and scale, and which build path suits your situation.
Why read this
The piece is practical guidance rather than a release announcement.
Business-driven design and conceptual modeling
Choose a build path
Implement and test iteratively
Design for query performance
Centralize security and standardize
Five groups of practices for building a performant semantic model, in the order the article presents them
The article argues that a semantic model should reflect business concepts rather than database schemas, and that conceptual modeling and stakeholder review should come before joins are configured. It describes three build paths: generating a model from natural language with Tableau Agent, generating one from existing CRM Analytics or Tableau Cloud data, and extending a verified model through composable modeling. It recommends starting from the Salesforce Customer 360 Semantic Model where applicable.
For performance, the guidance is to keep runtime joins to about six or seven, avoid many-to-many joins and expensive operators such as BETWEEN, join on primary keys, set explicit join cardinality, and align default filters with Data Lake Object partitioning. Secondary indexes and pushing heavy transformations upstream into Data 360 batch transforms or Calculated Insights are suggested for very large datasets. Security should be enforced through Data 360 Policies rather than model filters, which are meant for business relevance and can be bypassed by ad-hoc queries.
The piece also advises testing with the Test Model feature and visual checks for join fan-outs, and stress testing against production-scale volumes and concurrent query patterns before rollout.
You will learn what the Zero Copy v2 connector exposes in Snowflake, which pieces are generally available versus in preview, and how governance and private connectivity are handled across the two platforms.
Connect enterprise data
Give it shared meaning
Turn it into trusted context
Act on it in agents
Data moves from enterprise data to action in four stages, as the article frames Data 360's role
Salesforce describes Data 360 as a context layer for AI agents, and presents its Zero Copy partnership with Snowflake as the way to reach governed data where it already lives rather than copying it through ETL pipelines. The article reports more than 1,000 joint customers running workloads with native Zero Copy, and presents this as the basis for Agent Context Engine, which combines structured data, unstructured content, knowledge graphs and agent memory.
The technical changes are the substance here. The v2 Zero Copy connector presents Data 360 objects as native Apache Iceberg tables inside a customer’s Snowflake account, queryable with SQL and usable with Streams and Dynamic Tables for near real-time change capture. OpenID Connect authentication secures access. Expanded Zero Copy with Horizon Catalog adds direct file federation with Snowflake-managed Iceberg tables, and Data 360 Private Connect Multicloud allows Data 360 on AWS to reach Snowflake on Azure without public internet traffic.
Availability is staged. The v2 connector and Private Connect for Snowflake on Azure are generally available, Expanded Zero Copy with Horizon Catalog and Snowflake Cortex Agents are in public preview, and the Salesforce and Informatica MCP servers in Cortex are in private preview. Tableau support for Snowflake semantic views is planned for general availability in early 2027.
You will learn what the announced Braze acquisition means for Marketing Cloud Engagement and Marketing Cloud Next, and which skills and architectural decisions it may affect.
Why read this
The article also makes clear that no combined roadmap exists yet.
Marketing Cloud Engagement
Relies on Journey Builder
Established automation offering
Marketing Cloud Next
Emphasises Data 360 and AI
Automation through Flow Builder
The two Marketing Cloud platforms the Braze deal must fit alongside, as the article describes them
Salesforce has agreed to acquire Braze, an engagement platform that reacts to customer behaviour in real time across email, push, in-app messaging, SMS and WhatsApp, in a deal valued at around $5.66B. Salesforce expects to close during fiscal 2028, subject to shareholder and regulatory approval. Braze says in an SEC filing that its integration with Data 360, Customer 360 and Agentforce remains a possibility and that no decisions have been made.
The article notes that Salesforce has not published a combined roadmap showing how Braze’s capabilities will overlap with Marketing Cloud Engagement, which relies on Journey Builder, and Marketing Cloud Next, which centres on Data 360, AI and Flow. Salesforce’s president of applications and marketing suggested Braze would extend Marketing Cloud Next with real-time, mobile-first engagement rather than replace Marketing Cloud.
For practitioners, the piece argues that the greatest risk is over-specialising in one orchestration tool, and that event tracking, identity, API integration and data architecture skills will grow in value. It also expects consulting work around integration, migration and platform rationalisation, while warning that pricing, packaging and third-party integration openness remain unknown.
You will learn the workflow for defining business metrics in OSI YAML, converting them to a Tableau SDM, validating and querying the model, and exposing it to Agentforce and MCP clients.
Why read this
The piece gives the concrete API commands needed to do this.
Ingest CRM data into DLOs
Define metrics in OSI YAML
Convert to SDM JSON
Create and validate the model
Query via Semantic Query API
The workflow runs from ingested data through the OSI specification to a queryable semantic model
The post argues that AI agents need a shared semantic layer so that terms such as revenue and pipeline are defined once and reused across reports, applications and agent answers. Data 360 ingests CRM records into data lake objects, and a Tableau semantic data model adds relationships, metrics and calculations on top of them. Business context is captured first in an OSI YAML specification, which the Salesforce converter translates into an initial SDM JSON file.
The article walks through creating, validating and querying the model with Salesforce APIs, including the converter command, the model creation and validation requests, and a semantic gateway query. It also covers business preferences that steer agent terminology and scope, and verified questions created through Q&A Calibration to make answers more consistent. Agentforce and Claude connected through MCP are shown answering the same pipeline question from the same model.
The post notes that governance through Data 360 classifications, tags and policies can limit access to sensitive data. Its conclusion recommends starting from one business question, defining its terms, validating results, and then extending the shared foundation to more questions and agents.
You will learn how Data 360 credits are consumed by rows processed, and how the choice between zero-copy, batch and streaming, and query or file federation changes that cost.
Why read this
You will also see when internal CRM pipelines are free.
Zero-copy
About 70 credits per 1 million rows
Best for infrequent or ad-hoc queries
Data must stay in external lake
Standard ingestion
500 to 2,000 credits per 1 million rows
Cheaper for frequent queries and transforms
Copies are acceptable
Zero-copy and standard ingestion differ in cost per million rows and in when each is cheaper
Data 360 credit consumption for many usage types is driven by the number of records processed or accessed, not by the rows returned. The article argues that pipelines which scan large volumes but return few rows waste credits, so architects should decide how data enters the platform before anything else. Data kept outside the platform, or filtered before ingestion, incurs no processing cost, and zero-copy avoids replicating data at all.
The article compares zero-copy with standard ingestion. Zero-copy suits infrequent or ad-hoc queries and data that must stay in an external lake, at roughly 70 credits per million rows processed plus compute on both sides. Standard ingestion, at 500 to 2,000 credits per million rows, is more economical when data is queried frequently and transformed. Streaming costs more than batch because of continuous micro-batching and index updates, so batch should be used unless near-real-time freshness is needed. Core CRM pipelines from Sales Cloud, Service Cloud, Marketing Cloud and B2C Commerce have consumed zero credits since August 2025, though some Industry Cloud connectors still incur standard usage.
For zero-copy, the article contrasts query federation over JDBC, which bills compute on both Data 360 and the external source and is best for datasets under 100 million rows, with file federation over Apache Iceberg, which reads storage directly, is faster on datasets over one billion rows, and has native change capture and Flow automation. It recommends enabling acceleration for frequently queried zero-copy data to avoid repeated external compute charges.
Readers will learn how to structure data, identity, and consent architectures before configuring journeys in Marketing Cloud.
Why read this
The piece outlines practical questions to answer during implementation planning to prevent scaling failures.
The article argues that Salesforce Marketing Cloud implementations often fail when treated as simple campaign setups rather than comprehensive data projects. It emphasizes that reliable personalization depends on a foundational layer that handles data trust, identity mapping, consent tracking, and system integration before any journeys are built.
It outlines key architectural decisions that must be made prior to configuration, such as identifying authoritative data sources, selecting shared attributes for decisioning, and determining refresh frequencies. The author warns against indiscriminately loading every available field into Data Extensions, recommending instead a focused dataset that supports cross-journey personalization.
Identity resolution and consent management are highlighted as critical components that require business-led decisions rather than pure platform configuration. While tools like Data 360 can consolidate records, organizations must define trusted identifiers, handle duplicates, and map household relationships. Consent must be modeled across channels, geographies, and purposes to ensure accurate suppression and compliance.
Finally, the piece advises designing integrations around business journeys rather than creating point-to-point connections between systems. It recommends evaluating implementation partners based on their ability to establish governance rules, reusable journey patterns, and clear data ownership, ensuring the platform remains sustainable as campaign volume grows.
You will learn how to collect execution telemetry without modifying application code and validate resource sizing changes against runtime and reliability metrics before production rollout.
The Salesforce engineering team addressed inefficient compute allocation across millions of daily Apache Spark jobs by shifting from success-based metrics to evidence-driven resource sizing. They identified that successful job completion often masks unused capacity, pod churn, and unnecessary executor allocations.
To gather comparable measurements without altering customer applications, the team instrumented a shared submission gateway using open-source listener hooks. These listeners exported structured telemetry through Kafka into a curated analytical layer, which standardized metrics for applications, jobs, stages, and executors while explicitly tracking missing data and retry semantics.
The team used the modeled data to distinguish between genuine waste and underlying bottlenecks like shuffle failures or memory pressure. By testing targeted reductions in initial executors and maximum limits, they validated changes against p95 runtime, spill rates, and platform health indicators. The adjustments yielded a thirty-one percent weekly compute cost reduction without degrading performance or increasing failures.
Salesforce Developers BlogThe guide demonstrates how structure-aware splitting preserves table headers and speaker labels, resulting in more accurate Agentforce responses.
You will learn how to deploy Python-based Code Extension functions to override default chunking in Data 360 search indexes.
Why read this
The guide demonstrates how structure-aware splitting preserves table headers and speaker labels, resulting in more accurate Agentforce responses.
The article argues that poor chunking, not the language model, is often the real cause when an Agentforce agent misses details from a PDF or misreads a conversation, because retrieved snippets lose their context before the model sees them. Native Data 360 chunking (section-aware and semantic passage extraction) works well for standard content, but it struggles with complex multipage PDFs, dense tables and multispeaker transcripts when character or token-based splitters are used. Code Extension functions let developers deploy sandboxed, stateless Python chunking logic inside the Salesforce trust boundary, authored with the Data Custom Code SDK, deployed via the Salesforce CLI or UI, and wired into search index creation. The authors caution that their regex and pattern-based examples are illustrative happy-path code, not production-grade, and can produce false positives and negatives.
The piece tests three scenarios, comparing native chunks with custom ones. In a quarterly financial report, native splitting left numbers stranded without their column headers, making a growth-rate question unanswerable; a structure-aware function that detects table boundaries and repeats headers in sub-chunks lets the agent give a precise segment comparison. In a recorded escalation call with five speakers, native chunking produced unattributed walls of text, while a sliding-window function over speaker turns (window of 6 turns, stride of 3, capped at 1,500 characters) kept questions with their answers and preserved speaker names and timestamps, so the agent could identify a contractual downtime commitment. In a regulated-sector document, native chunking exposed SSNs, card numbers, emails, phone numbers, medical record numbers and other identifiers in the vector index; a regex pass before chunking masked the standard formats, though proprietary identifiers such as medical record, payer member, case and token numbers passed through until organization-specific patterns were added. The conclusion is that Code Extension is an escalation path for complex data rather than a replacement for native chunking, and it points developers to the documentation, the Code Extension setup in Data 360 and a video playlist to get started.
By Chandan Agarwal · Salesforce Developers BlogThe guide also covers connecting these scripts to batch data transforms and monitoring execution through the platform.
You will learn how to structure, validate, and deploy Python scripts as managed Code Extensions within Data 360.
Why read this
The guide also covers connecting these scripts to batch data transforms and monitoring execution through the platform.
Northstar Outfitters faced inconsistent product data across commerce systems and needed a trusted catalog. Native visual transforms could not handle the required standardization, enrichment, and scoring logic, prompting the use of Data 360 Code Extension to run modular Python and PySpark scripts directly on Salesforce infrastructure.
The solution relies on isolated runtime environments where scripts read from and write to designated Data Lake Objects using explicit permissions defined in a configuration file. Developers validate logic locally against sampled data before packaging the script, dependencies, and tests into a deployable archive.
The guide details both Setup interface and Salesforce CLI workflows for uploading packages, selecting compute sizes, and linking the extension to a batch data transform. Execution, scheduling, and logging are handled by the platform, while operators can trigger runs and review history through REST API calls.
By Scott Nyberg · Salesforce EngineeringThe piece also explains how to maintain consistent identity resolution and dynamic native rendering across iOS, Android, React Native, and Flutter using server-side decisioning.
You will learn how to separate app instrumentation from campaign configuration so marketers can update experiences without app releases.
Why read this
The piece also explains how to maintain consistent identity resolution and dynamic native rendering across iOS, Android, React Native, and Flutter using server-side decisioning.
The team built a unified architecture to deliver real-time mobile personalization across iOS, Android, React Native, and Flutter. They solved platform fragmentation by establishing consistent schemas and event semantics while allowing each technology stack to use its native rendering capabilities. A bridge layer enables React Native and Flutter to access native APIs through iOS and Android, preserving a single SDK interface for developers.
Developers define stable content zones and register approved native components once. Marketers then configure templates, targeting, and component selection through the Salesforce user interface. The platform serves this metadata via a global CDN, and the mobile SDK applies the rules to dynamically render the correct native element without requiring new app builds.
Real-time identity resolution relies on Data 360 to merge anonymous web, email, and mobile interactions into a single profile during authentication. Decisioning runs entirely on the server, so the SDK only receives final experience instructions rather than raw customer data. Teams can validate targeting and layout using a QR code preview flow that injects live profile attributes into the simulator before launch.
By Raveendrnathan Loganathan and Tobias Muehlbauer · Salesforce EngineeringThe piece also explains how Data 360 separates runtime state from durable memory to maintain governance across models and channels.
Readers will understand how the Agent Context Engine retrieves, reconciles, and packages authorized evidence into token-efficient Context Packs for AI agents.
Why read this
The piece also explains how Data 360 separates runtime state from durable memory to maintain governance across models and channels.
Salesforce Engineering describes how Data 360 assembles governed, authorized context for AI agents without loading an entire enterprise data estate into every prompt. The core challenge is identifying the requester, purpose, authority and freshness needs, retrieving evidence from structured and unstructured sources, reconciling conflicts, and applying access, masking, residency, consent and purpose controls, all within token, latency, freshness and cost budgets. Data 360 serves as a shared runtime foundation across Salesforce, platforms such as Databricks and Snowflake, operational systems and knowledge bases, with data kept authoritative where it lives through ingestion and Zero Copy. The article cites Q2 FY27 production usage of 104.3 trillion connected records, including 81.5 trillion Zero Copy rows.
The Agent Context Engine works in two directions. On the outbound path it moves through six runtime stages: Resolve (identify requester, agent, purpose, authority, entities and freshness), Plan (choose access paths such as relational, search, graph, RAG, streaming and federated), Reconcile (rank evidence while keeping conflicts visible), Govern (apply controls before retrieval or action), Compile (fit evidence, working state, memory and citations into a token-fit Context Pack and action contract), and Learn selectively (return typed interaction traces and decide what to promote, update or expire). On the inbound path, conversations, tool calls, decisions and outcomes return as traces for audit and evaluation. Only approved facts move into durable memory, and only validated outcomes become learnings.
Data 360 separates runtime state into four forms: working context for the current reasoning step, short-term memory for the active trace across turns, long-term memory for approved facts and commitments that belong to the enterprise, and learnings distilled from validated outcomes and corrections. The article stresses that a transcript is not automatically enterprise memory. Trusted Context sits within Salesforce’s Enterprise AI Harness, alongside Trusted Models, Agency, Actions, Governance and Security, and the same governed context reaches Agentforce, custom and partner agents, and SaaS assistants through SDKs, API adapters, MCP and APIs. The authors also report performance results across production retrieval, high-concurrency queries, analytical SQL, ClickBench, TPC-derived and LDBC graph workloads, while noting these are not directly comparable. They propose an Agentic Context Benchmark, which would count a result only when the resulting action is both authorized and completed. New capabilities are planned to begin rolling out in early FY28, with availability varying by product, region and agreement.
By Scott Nyberg · Salesforce EngineeringYou will also see the architectural tradeoffs used to achieve sub-two-hundred millisecond response times for Agentforce interactions.
Readers will learn how the engineering team partitions and indexes Data 360 data graphs to resolve complex customer identities while maintaining strict data isolation.
Why read this
You will also see the architectural tradeoffs used to achieve sub-two-hundred millisecond response times for Agentforce interactions.
The engineering team built Data 360 data graphs to solve the context gap for Agentforce by unifying fragmented customer data across accounts, entitlements, and products. Instead of running multiple queries at runtime, the system pre-aggregates relationships into cohesive data products that agents can call directly.
Resolving customer identity required a partitioned architecture that keeps the broader identity graph separate from customer success views. This ensures that prospect data and information belonging to different tenants remain strictly isolated while still supporting complex many-to-many relationships.
Performance was optimized by analyzing access patterns, designing smaller multi-graphs, and building targeted indices to avoid full table scans. The team shifted from static data models to flexible structures that support semantic search and unpredictable agent queries, ultimately achieving median response times under two hundred milliseconds without dedicated autoscaling infrastructure.
It explains how Agentforce Coworker fits into the AIforce stack, how its routing and security model work, and gives a concrete setup path via Data 360 and Salesforce Go.
Agentforce Coworker is presented as one of three main pillars of AIforce, Salesforce’s interface layer announced at Dreamforce ’26. Its purpose is to bring Salesforce data and agents to where users already work, such as Lightning Experience, Slack, Microsoft Teams or AI tools, instead of making them switch between applications. It targets two problems: tab-switching between many tools, and users struggling to know which agent to use for a task as agentic workforces grow. It acts as a router agent, the front door to an org’s agents, sending queries to the right one. Its foundations are Agentforce (the digital workforce layer it orchestrates), Customer 360 (CRM as the unification point for customer data) and Data 360 (the grounding layer that indexes and governs data for AI).
The piece describes three interaction modes. Find lets users ask questions in plain language and get contextualized answers drawn from CRM data, Slack, files, knowledge bases and other connected sources. Catch Up proactively surfaces changes in the user’s scoped workspace, such as what they missed while in meetings. Plan and Act takes outcome-level instructions and routes the work to the appropriate agent. Security relies on the existing Salesforce model: profiles, Permission Sets and Sharing Rules still apply, so Coworker only sees what the user is allowed to see, following least privilege.
The author’s hands-on setup path begins with confirming Data 360 is enabled, checked under Data Cloud Setup Home, and assigning the Agentforce Coworker Admin Permission Set to administrators. Coworker is then switched on through Salesforce Go by searching for it in Setup. Administrators choose data sources under Manage Data, with CRM data enabled by default. Finally, users need the Agentforce Coworker User Permission Set, which can be assigned via Manage Access to Agentforce Coworker in Manage Users. In use, the Ask button beside global search opens the Ask: Home tab, and the author recommends starting a fresh session for unrelated prompts to avoid carrying over context.
The article closes with caveats. Data 360 must be enabled before Coworker works, though not every data set must be shared with it. Coworker is available in Enterprise, Unlimited and Agentforce 1 Editions. Billing is either consumption-based through Flex Credits or seat-based with usage caps, so costs can rise quickly and the author suggests monitoring value with tools such as Agentforce Grid and Agentforce Observability. The author also advises planning and cleaning data and permissions before rollout.