Ai.
All posts.
Getting an MCP Server Through ChatGPT Plugin Review: Two Rejections and What Fixed Them
OpenAI approved OrgPulse for ChatGPT after two sign-in rejections. Reviewer accounts, new-device codes and the credentials form explained, for MCP builders.
Agentforce and Data 360 Access Control: What Your Agent Can Retrieve, and Who Decides
Many Data 360 orgs run an Allow All policy, record-level policies may miss your agent user, and Trust Layer masking is off for agents. What that means.
ai-ready-sf-repo: Why Our Claude Code Framework for Salesforce Runs on Evidence, Named Owners and Guardrails in Code
Why ai-ready-sf-repo makes every AI claim cite a file, every decision name an owner, and every rule live in code, and what one real Salesforce story taught us.
Connecting ChatGPT and Claude to Salesforce Without Sharing Credentials: A Read-Only Integration User Behind an MCP Server
How OrgPulse connects Claude and ChatGPT to Salesforce over MCP: a packaged External Client App, a read-only integration user on JWT bearer, no admin token.
Intent → Spec → Test on Salesforce: Where Each Layer Lives, From Sandbox to Production
How to apply Intent → Spec → Test on Salesforce: where must-nevers, rules and tests live, which tests run in which org, and how to guard risky metadata.
Anthropic's Scion vs Git: Why Anthropic Replaced Git, and What It Means for Your Team
Anthropic replaced Git in its monorepo with Scion after AI agents pushed merges past 30,000 a day. How Scion differs from Git, and what your team can borrow.
Intent → Spec → Test: Linking Must-Nevers to Held-Out Tests, and When Not To
How to link must-nevers to rules and hidden tests by ID, keep those tests out of an AI agent's reach, and use the model only where a mistake would be costly.
CLI vs MCP vs Purpose-Built Tools, and Skills vs Agents: Claude Code, Copilot and Agentforce Vibes Compared
How CLI, MCP and purpose-built tools compare on token cost, what skills and agents are, whether agents can use skills, and how Claude Code and Copilot differ.
Salesforce in Claude: Token Cost of OrgPulse vs DX MCP vs the sf CLI
We asked Claude three Salesforce admin questions through OrgPulse, the DX MCP Server and the sf CLI on three models, and measured tokens, cost and accuracy.
AI-Assisted Delivery Methodologies Compared: Which One Fits Your Team?
Vibe coding, spec-driven development, AWS AI-DLC, test-first agents and more, compared side by side, with pros, cons and the team structures each one fits.
Archaeological Specs: Using AI Agents Safely on a Salesforce Org You Didn't Build
Before an AI agent changes an org you didn't build, map what it does now: retrieve metadata, trace dependencies, write an as-is spec and pin it with tests.
Jev vs Clef vs Laya: Cloudflare's Clef Decision Model Tested on 600 Business Decisions
I ran Cloudflare's new Clef and Clef-flash through the same tests as Jev, Laya and Einstein: 600 decisions plus 7 real failures. Accuracy, speed and cost.
SalesBleed: How a Web-to-Lead Form Got Agentforce to Leak CRM Data, and What to Change in Your Org
SalesBleed slipped past the Trusted URL fix from ForcedLeak and used Slack link previews to leak CRM data. How it worked and what Agentforce settings to change.
Salesforce Case Triage With Decision Models: Jev and Laya Against an LLM, Measured
I built one Salesforce case triage process three ways: Einstein, Jev and a self-hosted Laya, on the same 300 cases. Accuracy, speed, cost, and what broke.
Why Did My Salesforce Deployment Fail? Letting Jev, Laya and Einstein Read the Error, Measured
Can AI read a failed Salesforce deployment and say what broke and who should fix it? I tested Jev, Laya and Einstein on 300 test failures and 7 real ones.
Post-ForcedLeak: Hardening Agentforce Against Prompt Injection (The Setup Steps Nobody Published)
ForcedLeak turned a $5 domain into an Agentforce data-exfiltration channel. The concrete Setup hardening steps to shrink your prompt-injection blast radius.
The Specification Crisis: How Different User Personas Complicate Software Development
Explore how different user personas in software development teams create specification drift, amplified by LLMs, and discover solutions for better collaboration.