The Blog
Practical Salesforce engineering – Apex patterns, integration architecture, and platform decisions explained.
-
Claude inside Salesforce: what Agentforce’s partnership with Anthropic actually enables
What Salesforce’s Anthropic partnership really changes in Agentforce, and how a technical lead decides when an action should actually use Claude.
-
Identity resolution in Data Cloud: the unglamorous work every agent depends on
Agentforce can’t act on a customer it can’t recognise. Here’s how Data Cloud identity resolution actually works, and where it quietly breaks.
-
Koa and Salesforce’s own reasoning model: what it means for agent cost and latency
Salesforce’s native reasoning model changes the cost and latency math for Agentforce. Here’s when to use it, and when to still call Claude instead.
-
Long-running agents: what changes when Agentforce pursues a goal over days, not one request
Agentforce agents that run for days, not minutes, need real state design: durable goal records, timeouts, idempotent actions and goal-level observability.
-
Job-ready agents like Casey and Hunter: evaluating a prebuilt agent against a custom build
Casey and Hunter promise a job-ready Agentforce agent out of the box. Here is the test I actually run before deciding that beats a custom build.
-
AgentExchange and the rise of prebuilt agents: build, buy, or customise?
AgentExchange makes installing a prebuilt agent easy. Knowing whether to buy, build, or customise one for your org is the call that actually matters.
-
Hosted MCP servers on Salesforce: what they are and why they matter for integrations
What Salesforce hosted MCP servers actually expose, how they differ from External Services, and where they genuinely fit in a real integration stack.
-
Agent Script and the Agentforce Vibes IDE: what changes for Salesforce developers
Agent Script turns Agentforce logic into reviewable text instead of click-based guesswork. Heres what actually changes for Salesforce developers.
-
Testing AI agents before production: what “agent QA” actually looks like on Salesforce
Apex coverage won’t catch a bad agent decision. Here’s what to actually test before Agentforce goes live in a production contact centre org.