A while back I posted about five ways I use AI inside our sales motion, and it hit a nerve. The post pulled 302 reactions and 145 comments, and almost all of them were a version of the same question: which tools do I actually still pay for. I run Growth Cab, the GTM advisory I founded, and I sell every week, so this is not a demo I watched once and forgot. It is the research and prep layer I now run before nearly every call. The honest headline is that AI did not replace my sales skills. It deleted the two hours of manual research that used to sit in front of every good conversation.
Most of the noise about AI for sales is about automation that fires more emails at more strangers. That is the part I care about least. The version that changed my week is quieter. It reads the company, the person, the trigger event, and the tech stack faster than I can open the tabs, and it hands me the one thing that makes a call land: context the buyer did not expect me to have. Below are the five workflows I kept, and the two spots where the whole thing still breaks.
in
“5 Perplexity Computer workflows that replace your entire GTM stack. Most GTM teams are drowning in tools: Instantly, Apollo, Clay. They are expensive, slow to learn, full of gaps.”
What AI for Sales Actually Replaces
The tools most GTM teams stack together, the Apollos and Clays and sequencers, each solve one slice of the job and charge for it, and for a lean team learning them eats more hours than they save. What I wanted was not another tool to learn. I wanted to collapse the research and prep those tools scatter across five tabs into one place I can ask questions in plain language. That is the real promise of AI for sales for a founder or a small team: fewer seats, less glue work, and a single surface where prep actually happens instead of getting skipped because the day got busy.
The Five Workflows I Run Before Every Call
The first is call prep. Before a meeting with a cybersecurity company last month, I asked for a briefing and it surfaced a recent executive hire and a positioning change that were nowhere on their website. I walked in already knowing what was eating their budget, with no thirty minute scramble beforehand. Prep used to be the first thing I cut when the calendar got tight. Now it costs me four minutes, so I never skip it.
The second is prospect research. When I needed to understand a VP of Sales before reaching out, it built a full profile: background, recent company news, and a trigger event I could actually use. That trigger became my opening line, which beats any template. The gap between a cold message and a warm one is usually one specific detail, and this is the fastest way I have found to get that detail.
The third is demo mapping. Before a demo, I have it map our offering against the account's existing stack and stated priorities, so I can tailor the conversation to their real situation instead of running a generic walkthrough. A demo that speaks to the buyer's actual setup lands very differently from one that shows every feature and hopes something sticks.
The fourth is outreach drafting. When I write a first message, it drafts personalized opening lines off real signals like recent hires, funding rounds, and product launches, and I refine from there instead of staring at an empty box. I do not send what it writes untouched. It gets me to eighty percent, so my time goes into the judgment part, the last twenty.
The fifth is competitive intelligence. One competitor kept coming up on calls, so I had it track their launches and positioning. The next time a prospect raised them, I explained exactly where we were different and handled the objection on the spot instead of promising to follow up. Losing the room to I will get back to you is one of the quiet ways a live deal goes cold.
Where AI for Sales Still Breaks
Now the honest part, because those five workflows are not the whole story. AI for sales fails in two places, and I have watched people lean on it straight into a worse call. The first is trusting the output without checking it. It will state a funding round or a job title with total confidence and be a quarter out of date, and if you open with a wrong fact you have done more damage than if you had done no research at all. I treat everything it gives me as a strong first draft that still needs one human check before it reaches a buyer.
The second is expecting the research to do the selling. It does not. It gets you into the room prepared, and then the call is still yours to run: the listening, the framing, the read on whether this person can actually move a deal. AI for sales removes the grunt work in front of the conversation. It does nothing for the conversation itself, and teams that expect it to close for them end up with beautifully researched calls that still go nowhere.
Turning It Into a Repeatable System
The reason this is worth writing down is that it repeats. Five prompts, run in the same order, before every meaningful call: prep the company, research the person, map the demo, draft the outreach, track the competitor. Once the sequence is a habit, the prep tax that used to make me skip research on busy days is simply gone, and the floor on my calls rises because every one of them starts from real context instead of a blank page.
Write your five prompts down and save them where your team can run them, because the value is not one clever query. It is that anyone selling for you can walk into a call as prepared as you would be. That is the version of AI for sales that actually compounds. It becomes a prep system your company owns rather than a tool you rent by the seat.
I break down one revenue play like this every morning in The Revenue AI Brief, my daily newsletter for founders and GTM teams who want systems instead of scattered tips. If this was useful, the original post with all five workflows lives on my LinkedIn, where I share what is working inside Growth Cab in real time.
Disclosure: the workflows above run on Perplexity Computer, and I partner with Perplexity. I only write up tools I actually use in our own sales motion, and these five are the ones I run for real.
