Top sales reps use AI on the work around selling rather than on selling itself. A weekly workflow scans target accounts for new contacts and company changes, then drafts personalized outreach into a sheet the rep reviews before anything is sent. After calls, AI turns transcripts into a one-page summary a champion can forward, and deal updates come through chat while the CRM stays the source of truth.
Key takeaways
- Prospecting runs every Monday at 8am. A workflow looks for former Salesforce users who have joined or been promoted at target accounts, and for company changes worth acting on, then drafts outreach.
- Nothing goes out unreviewed. Drafts land in a Google Sheet where the rep edits, approves or omits each one.
- The CRM is headless. Salesforce holds the data, but the rep reads and updates it through Claude, Slack and Slackbot.
- One page moves the deal. After each call the transcript becomes a one-page PDF of challenges, examples and next steps that a champion forwards internally.
- Quotas have not changed yet. Salesforce has not adjusted quotas or territory sizes for AI productivity gains, and RevOps will need to separate real productivity from more activity.
There’s a lot of discussion about how AI is going to change sales, but I’m more interested in what good salespeople are actually doing with it today.
I recently sat down with Vicky Vo, a commercial AE at Salesforce, for our latest Deep Dive. Vicky and I worked together at Procore, where she started as a BDR straight out of college. She’s spent more than a decade selling to contractors and field service businesses, and is now focused on winning new customers across industries including construction, staffing and professional services.
What I found interesting wasn’t the number of AI tools she’s using, but how she’s connected them to the work she already does.
Her Monday morning prospecting is largely done before she starts work
As a new logo AE, Vicky manages a large number of accounts, many of which Salesforce has been prospecting for years. Historically, keeping track of who to contact, when to follow up and what had changed at an account was largely manual.
Now she has a workflow that runs every Monday at 8am, looking for two things:
- People who might open a door. Someone who used Salesforce at a previous company and has recently joined or been promoted at one of her target accounts.
- Company changes worth acting on. Acquisitions, growth or other events that create a reason to reach out.
When something relevant comes up, Claude drafts personalized outreach based on the account, the event and the contact.
The part I particularly liked is that the emails don’t go straight out. They land in a Google Sheet where Vicky can review, edit, approve or omit them. She gets the benefit of doing research and drafting at scale, without losing control of what goes to customers.
And she built this herself. There wasn’t a formal rollout or a prescribed workflow. She experimented until she found something that worked, and now shares those learnings with the team.
Salesforce is still the source of truth, but she doesn’t need to work in it
We’ve spent years in RevOps trying to get sellers to work in the CRM and reduce the number of tools they jump between.
Vicky described something quite different. She can ask Claude for a summary of her opportunities, pull deal updates through Slack, and use Slackbot to complete tasks in Salesforce.
She described Salesforce’s internal approach as “headless.” The CRM still holds the data, but the rep doesn’t need to be in Salesforce to access or update it.
For RevOps, I think this is an important shift. We’ve historically spent a lot of time thinking about the interface and how to get reps to use it. Increasingly, the question will be whether the underlying data and processes are good enough for AI to work across them.
She’s using AI to help move deals forward, not just prospect
Another example I liked was how Vicky handles customer meetings.
After a call, she uses the transcript and notes to create a one page PDF covering the customer’s challenges, what they discussed, relevant examples and next steps.
It’s something her champion can forward internally to bring other stakeholders into the conversation. Simple, but useful when you’re selling into organizations where several people need to agree on a purchase. We built the same idea into a client’s own stack, with three Claude skills that turn Salesforce and Gong data into a mutual action plan, an ROI case and a proposal.
Internally, each deal has its own Slack channel, where call summaries, actions and deal history are captured. That context carries over when an account changes hands.
She also shared a good example of how her team is learning from each other. One of her colleagues spent time refining a prompt for executive outreach. Once he had the output right, he asked Claude to consolidate his instructions into a reusable prompt and shared it with the team.
Rather than everyone starting from scratch, they can build on what someone else has already figured out.
Not every customer is living in the same AI world
One of the more interesting parts of our conversation was how different Vicky’s customers are from the technology companies selling to them.
Many are family run businesses that have been operating successfully for decades, sometimes with very little technology. They’re proud of what they’ve built, and understandably cautious about changing it.
Her conversations aren’t necessarily about AI. They’re about where the business wants to go, what’s getting in the way and whether technology can help.
That’s a useful reminder that while we’re busy figuring out how to automate everything internally, many of the customers we’re selling to are at a very different stage.
What does this mean for sales productivity?
I asked Vicky whether Salesforce has started adjusting quotas or territory sizes to account for the productivity gains from AI. Not yet.
But I suspect this will become a much bigger conversation over the next year.
If a rep can research hundreds of accounts, identify buying signals and prepare personalized outreach in a fraction of the time, what does that mean for territory capacity, productivity expectations and ultimately headcount?
For RevOps leaders, I think the challenge will be separating genuine productivity improvements from simply producing more activity.
What I liked about Vicky’s approach is that she isn’t trying to automate selling. She’s using AI to take away a lot of the work that surrounds it, so she can spend more time with customers.
And she’s not waiting for someone in RevOps or IT to tell her how to do it.
That’s probably the biggest takeaway for me. Some of the most useful applications of AI in sales are being built by the people doing the job, not by the teams responsible for the systems.
Watch the full Deep Dive with Vicky.
Frequently asked questions
How can a sales rep use AI for prospecting?
Set up a workflow that runs on a fixed schedule, such as every Monday morning, and scans target accounts for two signals: contacts who used your product at a previous company and have joined or been promoted at a target account, and company changes such as acquisitions or growth. AI then drafts personalized outreach for each signal, based on the account, the event and the contact.
How do you keep AI-drafted sales emails personalized?
Keep a person between the draft and the send. Drafts land in a shared sheet where the rep reviews, edits, approves or omits each one before anything reaches a customer. The rep gets research and drafting at scale and keeps control of what goes out, which avoids the generic blast that AI outreach can turn into.
What is a headless CRM?
A headless CRM still holds the data and stays the single source of truth, but reps do not have to work inside its interface to read or update it. They ask an AI assistant for an opportunity summary, pull deal updates through a chat tool such as Slack, and complete CRM tasks from there. The question for RevOps shifts from interface adoption to whether the data and processes are good enough for AI to work across.
How can AI help move a deal forward after a sales call?
Use the call transcript and notes to produce a one-page PDF covering the customer’s challenges, what was discussed, relevant examples and next steps. The champion can forward it internally to bring other stakeholders into the conversation, which helps when several people need to agree on a purchase. A dedicated chat channel per deal keeps summaries, actions and history in one place when an account changes hands.
Should sales quotas change because of AI?
Not every company has adjusted yet, but the question is coming. If a rep can research hundreds of accounts, spot buying signals and prepare personalized outreach in a fraction of the time, territory capacity, productivity expectations and headcount plans all need another look. The work for RevOps leaders is to separate real productivity gains from more activity.