How to Start Using AI as a Sales Leader

Sales reps spend 75-80% of their time on admin rather than selling. That figure comes from Salesforce and HubSpot's own platform data - and it means most sales teams are running at a fraction of their potential output before AI enters the picture. For sales leaders trying to figure out where to start with AI, the answer is specific: prospect research, personalised outreach, and call analysis. These three areas deliver the fastest time savings and the most immediate improvement in rep quality.

Why sales leaders need to adopt AI now

Most sales leaders already know they're behind. The embarrassment of not having figured out AI yet is real - and Richard Smith, Head of Growth at MySalesCoach, names it directly: "You're probably already being left behind. Hence why you're possibly on this webinar today - because you're sat there thinking, I probably feel like I'm not capitalising on this."

That's not a scare tactic. A 2024 study found that 71% of employers now prioritise AI skills over experience when hiring salespeople. If a candidate can demonstrate AI aptitude - even at a basic level - they go to the top of the pile over someone with a longer track record who's still working manually. That stat was from last year. The number is almost certainly higher now.

Tom Ridley, Sales Coach at MySalesCoach with two decades in sales and sales leadership, frames the opportunity like this: getting 22% of your time back sounds abstract until you translate it. "Who here has ever said, 'I'd love to have an extra week in the month or a quarter in the year to close that deal'? That's what we're talking about."

But Tom adds a point most AI adoption conversations miss: time saved is only half of it. "There's no point increasing your selling time if you aren't actually getting better at selling. AI needs to help you be more researched, more timely, more relevant, sound more expert to your prospects - not just do more of what you were already doing."

How to use AI for prospect research

Most reps research prospects the same way they did five years ago. They go to LinkedIn, scroll the profile, maybe click through to the company website, look at the news tab - and after 15-20 minutes of that, they've often found nothing they can actually use. Rich describes what happens next: "Their frustration just leads them to cut corners. They jump on the call and maybe not show that they've done their research - because they actually spent 15-20 minutes and didn't find anything."

Multiply that across every cold call, every discovery call, every week - and the cost becomes significant. AI doesn't just make research faster. It finds what manual searching misses.

Tom's recommended starting point is Perplexity. Unlike general-purpose AI tools, Perplexity was built for research - it cites sources, scrapes full websites rather than single pages, and actively works to avoid hallucinations. The setup is simple: build a prompt that includes your value proposition, your ICP, and the specific outputs you need (company overview, recent news, hiring signals, technology indicators). Save that prompt inside a Perplexity Space with custom instructions. From there, research that used to take 20 minutes per account takes seconds.

For teams with a larger prospect list, Tom demonstrated pulling that same prompt into a Google Sheet via API - researching hundreds of accounts simultaneously, with the output appearing directly in the cells. No manual steps required.

The prompt structure matters. "The most common mistake," Tom says, "is people just ask the AI to do research on a company - but they don't give it context. They don't tell it what their business does, what they sell, what their ICP looks like. The AI can't know what 'good research' means for you without that."

How to use AI for personalised outreach

Personalisation is where most reps say the right things but don't do them. The manual version - scrolling through a prospect's LinkedIn posts looking for something relevant to reference, checking earnings reports, reading hiring pages - takes 10 minutes per contact on a good day, and most reps don't do it consistently because it doesn't scale.

AI changes the economics. The research gathered in Perplexity becomes the input for a personalisation engine - fed into a ChatGPT project (or equivalent in Claude or Gemini) alongside your value proposition, your best-performing email examples, and your ICP profile. From there, outreach can be written that sounds like it took 30 minutes per email, because it's drawing on the same depth of context that 30 minutes of manual research would produce.

For teams that want a purpose-built tool, Tom highlighted two options at opposite ends of the market: Reggie.ai (enterprise-grade, integrates with CRM, Sharepoint, multiple data sources) and Tuesday (lighter touch, pulls from LinkedIn profiles directly, works well for high-volume prospecting).

Rich adds a point worth noting on authenticity: "Don't just use AI's output. Use it as - the big structure of the house is already built. You just want to add your own flavour to ensure you're remaining authentic and differentiated to your prospects." Tom agrees, and goes further: "AI is not the issue if you're getting people reacting negatively. The way you're prompting the AI is. If I removed the word AI and said a junior salesperson wrote this, we wouldn't label it that way. We'd just say the message isn't very good."

How to use call analysis to improve rep performance

Call transcripts are the most underused data asset in most sales teams. Tom is direct about this: "Call analysis - whether that's video or just the audio transcript - is probably the most valuable piece of data you can have in the sales function. It contains a wealth of data that up until AI was really difficult to get."

Most teams have access to some form of call transcription - Gong, Salesloft, Apollo's notetaker, or one of dozens of others. The problem isn't access. It's what teams do with the data afterwards. Rich describes what he sees most often: "They may still be watching back their calls one by one in full. Most people aren't even reflecting back on where they went wrong in their last call."

The two use cases Tom finds most valuable in practice:

Meeting prep slides. Take the transcript from the last call with a prospect, feed it into ChatGPT with the structure you want ("your situation, your challenges, the impact"), and it builds your "what I heard you say" slides in under a minute. "That exercise used to take me 30-45 minutes - essentially listening back to the entire call. It now takes me literally less than a minute. Makes me look super diligent and prepared."

Prospecting messaging. Feed multiple call transcripts into a project and ask AI to identify the most common problems prospects describe, in their own language. "That messaging becomes your best prospecting messaging - because it's coming from the voice of your prospect." This is how outbound teams can write cold outreach that sounds like it was written by someone who deeply understands the buyer's situation - because it was built from what buyers actually said.

For sales leaders specifically, call transcripts become a coaching tool. Instead of reading individual call summaries, feeding a batch of transcripts into an AI tool and asking for patterns - what's consistent across calls that aren't converting, where does talk time skew the wrong way, which objections are coming up repeatedly without resolution - gives leaders the data to coach at scale rather than one call at a time.

The Three-Layer AI Adoption Framework for Sales Leaders

MySalesCoach coaches sales leaders to think about AI adoption in three layers, each building on the previous one.

Layer 1: Time recovery

Start with tasks that are purely admin and have a direct AI replacement. Research is the clearest example - the work was always about gathering information, not about the manual process of gathering it. The goal here is reclaiming time without changing anything else about how the team sells. Prove the concept, show the time savings, build the habit.

Layer 2: Quality improvement

Once time is recovered, redirect it into quality. Personalisation is the obvious candidate - reps who had five minutes to research a prospect now have the research done in seconds, so they can spend five minutes actually thinking about how to use it. Call analysis at this layer means reps are reviewing their own calls and self-coaching between sessions rather than waiting for a manager to flag something.

Layer 3: Systematic intelligence

At this layer, AI is being used to generate insight that wasn't previously available at all. Feeding months of call transcripts into a project to identify pattern-level problems across the team. Using AI to do stakeholder profiling inside complex enterprise deals - understanding what each person said and did during a call, not just what the headline summary says. Using AI agents for real-time in-call guidance or next-call preparation based on the previous conversation. This is where the competitive gap opens up between teams that adopted early and teams that didn't.

Where to start if your team is starting from zero

Rich's advice for teams with no AI infrastructure yet: "Always experiment with something cheap or free first. Prove the concept. It might have some manual steps in it. It might feel a bit clunky. If you're using AI, there is 100% a way to automate it - but prove it works first."

The practical starting sequence Tom recommends:

1. Set up Perplexity (free tier is sufficient to start) and build one research prompt with your value proposition and ICP included. Use it for a week before every discovery call.
2. Create a ChatGPT project with your best email examples, your value proposition, and your ICP profile. Use the Perplexity research as input when writing outreach.
3. After your next three calls, feed the transcripts into ChatGPT and ask it to identify the most common problems your prospects described in their own words. Use that language in your next batch of outreach.

That sequence requires no new paid tools, no IT involvement, and no team-wide rollout. It's a proof of concept that one rep can run in a week.

Getting sales teams to consistently use AI across research, outreach, and call analysis - and then coaching them to improve the quality of what they do with the time saved - is one of the harder things to sustain as a sales leader. If you want your team working with a dedicated coach who has walked in their shoes and can build AI-enabled habits into regular 1:1 sessions, book a call with MySalesCoach to support your internal coaching efforts.

Listen to the full session with Tom Ridley and Rich Smith on the Surviving Sales Leadership episode page, or on Spotify and Apple Podcasts.

Frequently asked questions

What AI tools should sales reps use first?

Perplexity for prospect research, a ChatGPT or Claude project for personalised outreach, and whatever call transcription tool the team already has access to for post-call analysis. These three cover the areas where reps lose the most time and where AI delivers the fastest, most measurable improvement. Start with free tiers before buying anything new.

How much time can AI save a sales rep each week?

Salesforce and HubSpot platform data shows sales reps spend 75-80% of their time on admin rather than selling. AI applied consistently across research, outreach, and call analysis can recover around 22% of a rep's time - roughly equivalent to an extra week per month. The exact figure depends on how much manual work the rep is currently doing and how thoroughly AI is adopted across their workflow.

Will prospects be able to tell if outreach was written with AI?

Only if the prompting is poor. As Tom Ridley explains: "AI is not the issue if you're getting people reacting negatively - the way you're prompting it is." Outreach that's built on genuine research, fed with real prospect context, and then refined with the rep's own voice is indistinguishable from something written manually. The tell is generic AI output with no specific context - which is a prompting problem, not an AI problem.

How should sales leaders use call transcripts for coaching?

Feed batches of transcripts into an AI tool and look for patterns across calls - recurring objections, talk time distribution, where deals stall, what language prospects use to describe their problems. This gives sales leaders coaching data that previously required listening to every call individually. For individual rep coaching, ask the AI to identify specific moments in a call where the rep could have responded differently and why.

Is AI going to replace salespeople?

The data says no - but it will replace salespeople who don't use it. The 71% employer preference for AI skills over experience, cited by Rich Smith on the Surviving Sales Leadership session, points in one direction: AI competence is becoming a baseline expectation in sales hiring. Reps who can use AI to research better, personalise better, and learn faster from their calls will outperform those who can't. The human element - building trust, reading a room, navigating complex stakeholder relationships - doesn't get replaced. The admin does.