How Do You Brief an AI Assistant So It Actually Understands Your Business?
Last time, the audit. If you ran it, you already know whether you're getting generic answers because your questions are genuinely simple, or because you're re-explaining your business from scratch every time you open a new chat.
If it's the second one, here's the actual fix. Not a better prompt — a better way to brief.
Why "Just Ask It Better" Doesn't Work
The advice you'll find everywhere is "write a better prompt." Add more adjectives. Be more specific. Use the right keywords. Mimic the phrasing of whatever example got shared on social media last week.
That advice treats the problem as a wording issue. It isn't. The problem is structural: a general-purpose AI assistant starts every new conversation with nothing. No memory of your team size, your budget, your customers, or what you already tried last month. A better-worded question still starts from zero — it just starts from a more articulate zero.
Here's a way to see it clearly. Imagine hiring a freelance consultant for an afternoon, sitting them down, and asking one question with no other context: "How should I price my services?" A decent consultant wouldn't answer immediately. They'd ask what you do, who buys it, what you're currently charging, and what's prompting the question. If you skipped straight to demanding an answer with no context, you'd get a generic answer — not because the consultant is bad, but because you didn't give them anything to work with. An AI tool with no fixed role behaves the same way, except it won't stop to ask. It'll just answer with the average case and sound confident doing it.
What actually works is giving the assistant the same five things a genuinely useful advisor would ask for before giving you an answer.
The Five Things Every Useful Answer Actually Needs
We call this the FIRST ASK Framework™ — five parts, one for each letter: Facts, Intended Result, Resources, Sticking Points, Task to Start. It's not a prompt template you fill in line by line. It's the checklist of what's actually missing when an AI answer feels generic.
F — Facts. What's actually happening, and where does it currently stand? Not your whole business history — just what's relevant right now. What have you already tried? What already exists? If you've already attempted something and it didn't work, that's a fact worth including — otherwise you'll likely get the same recommendation again.
I — Intended Result. What does a useful outcome actually look like? Not "grow my business" — that's an activity wearing an outcome's clothes. What specifically needs to be different when this is done? "Post more on social media" is an activity. "Get enough inquiries that I'm not relying on word-of-mouth alone" is a result. The difference matters because an assistant can only aim at the second one.
R — Resources. What do you already have to work with — time, budget, skills, tools, people, existing content? This matters because most generic advice assumes you have less (or more) than you actually do, and recommends accordingly. A recommendation to "hire a part-time social media manager" doesn't help if that's not a resource you have — and a tool that knows that upfront won't waste your time suggesting it.
S — Sticking Points. What could get in the way? A deadline, a budget ceiling, a confidence gap, a platform limitation, a family or work constraint. This is the part people skip and then wonder why the advice doesn't survive contact with reality. If you only have evenings and weekends to work on this, that constraint should shape the answer, not get discovered halfway through trying to follow advice that assumed otherwise.
T — Task to Start. What's the actual first question, decision, or assignment you want help with — right now, not eventually? This keeps a broad situation from turning into an unfocused conversation that never quite lands anywhere useful.
The Audit: Run It On Your Own Question
Take a business question you've been sitting on — pricing, marketing, a hiring decision, anything real. Before you type it into anything, answer these on paper first:
What's actually happening right now? (Facts)
What does "solved" actually look like? (Intended Result)
What do you already have to work with? (Resources)
What could get in the way? (Sticking Points)
What's the first thing you actually need help with? (Task to Start)
Now ask the same question with those five things included, instead of the bare question alone. Compare the two answers side by side if you want to see the difference clearly — the bare question tends to produce a tidy, generic list. The briefed version tends to produce something that sounds like it's actually talking about your situation, because it is.
Here's the trap: most people either skip this because it feels like extra work, or they overcorrect and write a five-paragraph biography before every question. Neither is the point. The point is that a general assistant with no fixed role has to guess at all five of these every time — and it guesses with the average case, which usually isn't yours. Giving it even three of the five, briefly, changes the answer more than any amount of clever wording ever will.
A worked example, side by side. Bare question: "How should I price a new service?" A generic answer to that comes back with competitor-based pricing, a suggestion to "position in the middle," and a note about testing different price points — reasonable-sounding, and built for a business that isn't yours. Now the briefed version: "I currently charge $75/hour for this service and have three months of steady bookings at that rate (Facts). I want to know if I'm leaving money on the table without pricing myself out of my current client base (Intended Result). I have booking history and know my rough costs, but haven't done formal market research (Resources). I can't afford to lose more than one or two current clients over a price change (Sticking Points). I want a specific number to test with new clients first, before touching existing ones (Task to Start)." That's five sentences, not a biography — and the answer that comes back will actually be about your situation instead of a hypothetical one.
One more thing worth knowing: you don't need to run this full exercise every time. Once you've briefed a conversation properly, you can usually keep building on it without repeating everything — the problem is specifically when you're starting cold, in a new chat, with no context carried over.
This Is Already How Our Digital Team Members Work
This isn't a theory we're asking you to apply on your own. It's the actual behavior built into Remy, Pablo, Samantha, and Beau — our GLBS Digital Team Members. When you're unsure how to start a conversation with any of them, they're built to ask for what's missing from these five things, conversationally, only when it would actually change the answer — not as a rigid intake form, and not by making you repeat what you've already told them.
Here's roughly what that looks like in practice. Say you open a conversation with one of them and just say "help me with my marketing." That's a Task to Start with almost nothing else attached — no Facts, no Intended Result, no sense of your Resources or Sticking Points. Instead of guessing, the assistant will ask one or two questions aimed at whatever's most likely to change the answer: what you're currently doing for marketing, or what you want different by the end of this. It won't ask you ten questions before helping — it'll ask the one or two that actually matter, and then get to work with what it has.
That's the real difference between a blank chat window and a tool built with a fixed role: the role already knows what it needs to ask for, and it knows when it already has enough to just help you instead of interrogating you first.
A note on the name: the FIRST ASK Framework is something we built and are introducing here for the first time — it's new, and we're sharing it in that spirit rather than treating it as a finished, fully legally reviewed product name. Consider this the honest, working version, not a polished trademark rollout.
The Honest Ending
If you ran the audit and found you already instinctively give context before asking — good, you were probably already getting decent answers, and this post mostly confirms what you were already doing right.
If you found yourself skipping straight to the bare question every time, that's the habit worth breaking, whether you're talking to a general AI tool or to one of ours. It's a small change with a bigger effect than it looks like it should have.
Starting with the next post, we'll introduce the four GLBS Digital Team Members one at a time — what each one actually does, and how the FIRST ASK habit shows up differently depending on which one you're talking to. Two are free to try, with no purchase involved.
If you're not sure where the actual gaps are in how your business runs — beyond just AI conversations — that's worth a direct conversation rather than another guess.
