We started Green Orb in 1997, which is not a branding line so much as a calendar fact that occasionally makes me feel ancient in rooms where someone is pitching “the first real AI platform for mid-market.” I’ve sat through a dozen “first real” platforms over the years. A few were decent, some were fine for a narrow use case, and most of them were a sales process wearing a product costume.
What follows isn’t a manifesto or a list of life hacks. It’s the expensive patterns we keep seeing and the quieter habits that keep people out of ditches. Take what helps and ignore the rest.
Fundamentals are unfashionable, which is why they get skipped
Nobody gets excited about patching, and nobody wants a board slide titled “we verified the backups again,” yet monitoring, identity hygiene, restore tests, and a support path people will actually use prevent more damage than most “transformation” programs I’ve watched burn through cash. I don’t say that to be a killjoy. I say it because we’ve cleaned up after the other approach more times than I can count, where a new platform lands on top of an old mess and somehow the platform takes the blame while the mess that was always there goes unmentioned.
Before you fund novelty—AI agents included—ask a blunt question: are yesterday’s basics under control, or are we papering over them with a logo?
Overselling loves a vague requirement
When the business says “we need better security” or “we should do something with AI,” a lot of vendors hear an open checkbook. Ambiguity is not collaborative in those rooms; it is expensive. Precise outcomes shrink the space for rip-offs, and we’ve sat through meetings where three tools got proposed for one problem while the adjacent problem that actually ruins Friday afternoons got nothing at all. That isn’t strategy. That’s product attachment with a calendar invite.
Write the outcome first and then evaluate tools, and if a salesperson gets impatient with that order you’ve already learned something useful about the deal.
Second opinions are cheaper than rewrites
People treat outside review like distrust. I treat it like insurance. Independent review before a multi-year platform, a cloud migration, or an AI rollout almost always costs less than untangling lock-in you didn’t mean to sign. Years ago a client nearly bought a “suite” that bundled useful monitoring with four modules they would never staff. We asked what happened if they bought monitoring alone, got silence, then a different quote. That silence was the whole story.
AI multiplies whatever you already are
If you automate a clean process, you can scale value. If you automate a broken process, you scale the brokenness with better confidence language. Agents don’t fix ownership. They expose the lack of it at higher speed. Our AI consulting is deliberately unromantic: evaluate, select, implement, and govern, with permissions, human review, and a kill switch written down before anyone demos for the CEO. If that sounds boring, good. Boring is how you avoid becoming a cautionary anecdote someone else tells at a conference.
Contracts encode power, not friendship
Term length, auto-renewal, data portability, and who is allowed to touch the professional services decide who has leverage later. I’ve watched smart people skip page twelve because the demo was fun and the quarter was ending, and page twelve always comes back. Read the exit. If you can’t leave without a small crisis, you’re not a customer so much as inventory.
Calm still beats hype
The teams that avoid regret tend to sound less excited in meetings. They ask dull questions, they refuse urgency without evidence, and sometimes the highest-value thing we say is “not yet.” Nearly three decades in, I’m less interested in being right about the next wave and more interested in clients not funding someone else’s quota. Exceeding expectations, for us, looks like fewer irreversible mistakes rather than a longer shopping list.
If you want a partner who will tell you when not to buy, when to wait, and when a simpler path is stronger, that’s the work. It always was. The tools change. The incentives don’t unless you hire for independence on purpose. If this reads like a collection of scars, it is, because the alternative is pretending every year is year one, and neither we nor the businesses that pay for technology with real money have that luxury.