The independent daily for Salesforce admins, developers and architects

Curated by Abhinav (opens in a new tab)

Must-read

Six practices for taking Tableau MCP agents from prototype to production (opens in a new tab)

You will learn how to prepare a Tableau data model for agents, scope their tools and data, and build a test and monitoring harness before agents reach end users.

Why read this

The post draws on a regulated-industry customer deployment.

  1. Prepare the data model
  2. Define the agent's role
  3. Follow least privilege
  4. Load metadata into the prompt (optional)
  5. Pre-build queries for known questions
  6. Build evaluation and monitoring
Six practices for moving a Tableau MCP agent from prototype to production, in order

The post describes a large enterprise customer in a regulated industry that moved from Tableau dashboards to custom agents built on Tableau MCP. Off-the-shelf Tableau agents were too broad for the customer, who needed specialized workflows, agents that could act on results, embedding in a purpose-built application, and fine-grained control. Custom agents bring more need for testing, governance and monitoring, so the authors set out six practices.

The first is to prepare the data model. Business logic should live as close to the data as possible: calculated fields for fixed logic, Tableau Knowledge business preferences or field descriptions for logic shared across agents, and agent system instructions for logic specific to one agent. Teams should also remove duplicate fields, describe fields with units and caveats, name them in business terms, hide irrelevant fields and certify the data sources. The TDS Edit API, available in Tableau Cloud from October 2026 and Tableau Server 2026.3, lets teams add field descriptions in code.

The second is to define the agent’s role by answering what problem it solves, which domain it serves, what it can and cannot answer, and how it should respond. The answers shape its data, tools and system prompt. Tableau MCP now supports embedded data sources in 2026.3.

The third is to apply least privilege. The authors removed tools the agent did not need, starting with list-datasources, and pinned the agent to an approved data source ID. Tool scoping is set through environment variables on self-hosted deployments or site settings on hosted Tableau Cloud, and embedded data sources are scoped by project or tag.

The fourth, optional, practice is to load metadata into the system prompt at session start. This removed the get-datasource-metadata call and cut response time by almost 50 percent. The authors warn that this suits agents with one or two data sources and that stale metadata is a risk if schemas change often.

The fifth is to pre-build queries for frequent questions and expose them as custom tools, so the model only explains results. The authors advise against chasing perfect accuracy for open-ended questions, and suggest dashboards for figures that must be exact.

The sixth is an evaluation and monitoring framework: a question bank with correct answers, including refusals, run against each prompt, model and setting version, with LLM-as-a-judge grading for less deterministic answers, release-style versioning, production monitoring of latency, tokens and failures, and a regular review that adds new questions. The authors recommend an established agent observability platform, and they name LangSmith as an evaluation tool they used.

Read the original at Salesforce Blog (opens in a new tab)

Worth reading?

Recommended

Modelling FCA complaints data in Salesforce and avoiding the UTC time zone trap (opens in a new tab)

You will know how to model the fields COM001 needs for Cases, including a UK-local date that puts complaints in the correct half-year.

Why read this

You will also know how to test the setup before the 2027 deadline.

  1. Complaint received in UTC
  2. Convert to Europe/London date
  3. Store in Date_Received_UK__c
  4. Derive H1 or H2 formula
  5. Group report by legal entity
How the UK date feeds the half-year label used in COM001 reporting

From 1 January 2027, UK financial services firms will report complaints to the FCA in a new return, COM001, covering fixed half-years. The article sets out a data model for this built from standard Salesforce features, for orgs that hold complaints as Cases. Group reporting is removed, so each complaint needs a required legal entity, held as a restricted picklist or a lookup to a reference object that compliance controls. Every trading name and acquired brand must map to exactly one entity.

The return counts complaints from vulnerable customers and complaints about a failure to consider or respond to vulnerability, which can overlap. The author recommends two separate checkboxes rather than one picklist, with Field History Tracking switched on. Standard history is kept for 18 months, or 24 months via the API, unless Field Audit Trail is enabled, so it should not be relied on for longer retention.

The main pitfall is time zones. Datetime values are stored in UTC, so a complaint received just after midnight on 1 July during British Summer Time falls on 30 June in UTC and lands in the wrong half-year. The fix is a separate date field holding the UK date, set in a before-save trigger or invocable Apex using the Europe/London time zone, with the same done for the closed date. A formula on that date field then derives the half-year label, and reports are grouped by legal entity and that label.

The article also advises keeping the reason for a vulnerability marker in a separate field or related record, protected by field-level security, since health details can be special category data under UK GDPR. Before 1 January, it recommends testing a complaint created at 00:30 UK time on 1 July, running the same report twice a week apart, and confirming that changes to a vulnerability marker on a closed complaint are logged. The exact field names and validations are in the FCA’s COM001 Data Reference Guides.

Read the original at dev.to (opens in a new tab)

Worth reading?

Architecting HCP and HCO 360 data models and integrations in Life Sciences Cloud (opens in a new tab)

Readers will learn how to design the reference architecture, data model, and integration sequence required to implement HCP and HCO 360 profiles in Life Sciences Cloud.

Why read this

They will also understand best practices for managing affiliations, resolving duplicates, and aligning field territories.

This guide explains how to design and implement HCP 360 (healthcare professionals) and HCO 360 (healthcare organizations) in Salesforce Life Sciences Cloud. It starts from the problem that a single doctor can appear across several hospitals and in multiple systems such as CRM, marketing platforms, and external healthcare databases, leaving sales reps with inconsistent or incorrect information. HCP 360 gives a connected view of an individual professional’s details, workplaces, contact data, territory, and interaction history, while HCO 360 gives account managers a view of an organization’s location, departments, affiliated professionals, parent group, and engagement history.

The article recommends a reference architecture before any configuration begins. External data providers supply identifiers, specialties and addresses; master data management decides how records are matched and which system owns each attribute; Life Sciences Cloud runs customer engagement; and Salesforce Data 360 can unify sources for analytics and segmentation without automatically replacing the master data system. The data model should reuse standard Life Sciences Cloud entities such as provider, provider organization, affiliation, identifier, specialty, and territory before any Custom Objects are added. Affiliations must support multiple roles, dates and history, and HCO hierarchies must be kept distinct from professional affiliations.

For duplicates, the guide sets out a five-step process: standardize data, match on reliable identifiers, compare supporting attributes, send uncertain matches to a data steward, and keep source-system references. It warns against merging on name alone. For integration, it recommends loading HCOs first, then HCPs, then affiliations, then territory alignment, and finally validating counts, duplicates and relationships, using bulk processing and retry and error logging. The article also covers territory design beyond geography, keeping consent management separate from record-level security, and building HCP and HCO 360 pages on standard features first. Common mistakes include unnecessary Custom Objects, unclear data ownership, copying too much external data, overwriting affiliation history, and building interfaces before data quality is fixed. A pre-go-live checklist and an FAQ close the piece.

Read the original at SalesforceCodex (opens in a new tab)

Worth reading?