AI is quietly eating the entry-level work that builds judgment rep by rep. Why BrightHelm sponsors Dreamin' In Color and backs it with a real fellowship pipeline, not just a statement about the ladder.

Every Salesforce go-live has that moment: the process is live, users are logging in, and something small goes sideways a validation rule that's too strict, a Flow that behaves differently under real traffic. None of it is catastrophic, but all of it needs attention now, not next week. For years we handled that informally: "someone will keep an eye on things." But that's a hope, not a commitment. So we built Launch Watch your Project Manager, reserved exclusively for your go-live day.

When we saw the requirements for ACHIEVEability's fellowship no Salesforce certification yet, two days a week our honest first reaction was: is that enough runway? We said yes anyway. What made two days a week workable wasn't something we built for the fellowship; it was documentation designed to teach, plus Slackbot handling the repetitive onboarding scheduling so our human hours go to coaching. There's real anxiety about AI erasing entry-level tech jobs. We used AI specifically to invest more in someone just starting out.

When an AI-infrastructure founder got his GitHub account flagged, the exposure he'd been warning clients about stopped being theoretical so he moved his whole git stack onto infrastructure he controlled. Most organizations never get that wake-up call; they just carry the exposure until something forces the question. But "do we control our data?" isn't one question, it's three: Location, Mastery, and Use Rights. Here's the framework, why Use Rights is the sharpest edge in the AI era, and how BrightHelm runs every tool decision through it.

Our first part covered why Slackbot and Claude Tag are built for different jobs. This one is about the money and the pricing picture is more confusing than the vendor pages suggest. The 15-message-per-week Business+ limit that makes Slackbot a taste, not a tool; why real access means an Enterprise+ migration with lock-in in both directions; Claude Tag's different math when you're already paying for Claude; and three cost scenarios for a 50 person organization that genuinely land differently depending on where you start.

Slackbot and Claude Tag run on the same Claude model inside Slack but one is a personal agent and one is a team agent. The architectural difference that should decide which one you pick.

Salesforce's Informatica acquisition is a bet on data as core infrastructure. An Informatica World 2026 recap on why clean data is the non-negotiable precondition for agentic AI especially for NPSP-to-AFNP nonprofit migrations.

A Wharton study found AI can push accuracy below having no tool at all, while raising confidence either way. Why "cognitive surrender" is a system-design problem, and where in the build lifecycle deference turns catastrophic.

Earlier this year, Salesforce acquired Contentful. If you've been following this blog, you know what I think that means: a bet on Composable content architecture, and a signal that the structured-content-as-infrastructure era is here for enterprise buyers who are ready to do the work.

Every architecture taxes you, the only question is when the bill comes due. The monolith bills you downstream and later, often under duress when you're locked in. Composable bills you upstream and now, in the discipline of getting your structure right. AI agents have made that second choice newly urgent and they're quietly pushing whole organizations into it by accident. Here's how to trace the mechanism and choose on purpose, before the pull decides for you.

On June 1, Salesforce bought Contentful for a rumored $1B+ and skipped the cheap move of slapping an API on the content layer it already owned. That choice tells you what an AI agent actually needs: structured pieces it can recombine, not rendered blobs it has to parse. Here's what composable really means, why architecture and single-vendor procurement are two different decisions, and the honest answer on whether any of it reaches your nonprofit or public-sector desk yet.

There's a U.S. Army badge I've always loved: the Pathfinder, a winged torch carried by the troops who go in first, into unsecured ground, to light the way for everyone behind. First In, Last Out. It's also the cleanest name for how we work and why we just walked away from Jira, the industry standard, after running our actual business on it. A partner who'll tell you what to skip is worth more than one who only tells you what to buy.

The honest finale to our Slack Enterprise Grid series: Canvas sprawl, the hygiene habits that suddenly matter, and the management gaps to govern early because Slack's creation tools have outpaced its management tools.

Slackbot cleared half our automation backlog in a week. Part 2: why Skills make Flows obsolete, naming agents as governance (meet Sparks), and why Slackbot mirrors back exactly how clean your workspace really is.





Here's how a five-person company ended up on Slack Enterprise Grid, a product built for orgs with IT departments and thousands of users. The honest answer: we didn't migrate for the enterprise features. We did it for one thing, unlimited Slackbot, after Business+ rationed us to a genuinely useless 15 messages a week. Part 1 covers the 15-message trap, the migration itself, and why a tiny team runs a lot of channels because your channel model is Slackbot's map of your business.

We admire a plant's sunlit leaves but rarely the roots doing the unseen work below. Botany has names for both: phototropism, the visible reach toward light, and gravitropism, the quiet downward growth that anchors everything. For months I've been in a gravitropic phase, learning Salesforce Apex through late nights, Trailhead, and documentation, with progress measured in small wins nobody sees. Cutting that foundational phase short for quick, visible results leads to fragile growth. The groundwork isn't separate from progress. It is progress.

The Salesforce technical field is leveling, not because the work got easier, but because the moat got narrower. AI can write the formula and scaffold the flow. What it can't do is sit in a discovery call and notice the stakeholder flinching, or that two departments have been quietly fighting over data ownership for three years. That judgment work is now the load-bearing skill and the people who've done it longest, often unrecognized, are disproportionately women. That's why we sponsor WITness Success.

Your Salesforce SSO probably doesn't fully cover the 2026 MFA requirements. A practical guide to AMR/ACR signals, the privileged-user reporting gotcha, the report step-up surprise, and how to verify what your IdP is actually sending.

Two days after we published our analysis of Salesforce's 2026 security changes, Salesforce ran customer briefings and some of it confirmed our read, some added detail, and one piece pushed back on our framing. This post captures the deltas: why even SSO users with phishing-resistant MFA need a Salesforce-side method for report step-up, the step-up window flooring at two minutes, Experience Cloud's exemption from enforcement, recovery procedures you need before lockout, and the genuinely unsettled state of synced passkey support.

AI inference costs are falling faster than almost anything in computing history, so why are companies firing people now? Not because of what AI costs today, but because of what it's projected to cost in 2030. That's a bet, placed on someone else's livelihood. Running through Marx and William Morris, this essay argues there's a moral difference between the rote work that should be automated and the judgment-and-care work that shouldn't and that mission-driven orgs can use AI to reclaim work, not subtract people.