.webp)
Search for how to use AI in sales and most answers are long lists of ideas. Here are four workflows running at Vanderbuild or at our clients. For each one you get what starts it, what it produces, and what it changes.
Then comes the case where automation is a waste. After that, the two foundations you need first and the parts a person still owns.
How to use AI in sales: four workflows that work
1. A lead qualifier that reads every inbound request
Every request that reaches us, whether an email, a LinkedIn message, or a form on the site, goes through one agent. The agent qualifies it and filters out spam and irrelevant requests, such as SEO offers. It enriches the company data and writes the record into the CRM. Then it posts a note to our internal channel and archives the decision in a separate database. A qualified lead gets an email with a meeting proposal straight away.
The archive shows us what the agent let through and what it rejected, so we can correct the agent when it gets one wrong. The qualifier is the same setup we build for clients who want to qualify every inbound lead automatically.
The time saving is at least 10 minutes per lead. That is before you count the CRM entry and the enrichment. Multiply it by your monthly inbound and you have the business case. We describe a version of this flow step by step in our guide to an automated lead qualification system.
2. An offer copilot that works from the call transcript
Many of our offers are custom, so this one is a copilot: a person stays in the loop. After a sales call we sit down with the transcript and a set of frameworks. The agent drafts an offer built around what the client said on the call.
Time to offer used to be 3, 4, or 5 days. Now we send the offer within 24 hours, often the same day. The offer itself is a password-protected landing page made for that one client. For more on our AI sales setup, see how we use Claude for sales.
3. A reply agent on outbound campaigns
On outbound, the time to reply decides a lot. More than 20 minutes is already far too long, and ideally the reply goes out in the same minute. A reply agent handles that speed, and Salesforge has one built in. A salesperson still looks after the replies it handles. Answering fast gives you the best chance of getting a response.
4. Campaign agents for SDRs
At the top of the funnel, we teach SDRs and heads of sales at our clients how to deploy agents and connect data sources. With the right connectors and skills, they can create their own agents and prepare the campaigns they need. At the bottom of the funnel, we also have agents that help fine-tune contracts when an offer is more custom than the template.
AI in sales examples that work
Four workflows running at Vanderbuild or at our clients: what starts each one, what it produces, what it changes, and who stays in the loop.
Scroll the table sideways to see every column.
| Workflow | Trigger | Output | What it changes | Person in the loop |
|---|---|---|---|---|
| 01 Lead qualifier | An email, LinkedIn message, or form request reaches us | Qualified or rejected lead, CRM record with enriched data, a note in our channel, and a meeting proposal email | At least 10 minutes saved per lead | We review the archive of every decision and correct the agent when needed |
| 02 Offer copilot | A sales call ends and the transcript is ready | A draft offer built on what the client said, published as a password-protected landing page | Time to offer down from 3 to 5 days to 24 hours at most | We prepare the offer with the agent; a person stays in the loop |
| 03 Reply agent | A prospect replies to an outbound campaign, for example in Salesforge | A reply sent within minutes, ideally in the same minute | Faster replies; more than 20 minutes is already far too long | A salesperson looks after the replies |
| 04 Campaign agents for SDRs | An SDR or head of sales needs a campaign | Agents the SDRs create themselves from connectors and skills | SDRs and heads of sales prepare the campaigns they need | SDRs and heads of sales, trained by us |
Before a workflow runs fully on its own, a person controls it.
When using AI in sales is a waste of time
We do not automate prospecting when the total addressable market is small. If you sell to the 10, 20, or even 50 largest companies in one country, work account by account.
Don't waste time on automation there. Pick up the phone, send that email, do proper research, and give the eight-hour day to being as relevant as possible and knowing everything about that company.
No tool replaces how good you are as a salesperson or as someone who builds relationships. Everything else, every task you repeat by hand, can be automated, especially now.
How to use AI in sales on two foundations: knowledge and process
Two foundations come first: knowledge and process.
Knowledge. AI works on context, so it needs one source of truth for each functional area. That means sales, marketing, HR, legal, finance, administration, and general company knowledge. That is the structure we use ourselves. The knowledge is stored in the cloud, in Google Workspace, Outlook, or your internal systems. It is sorted by the area it describes and keeps flowing between people. We call that compressed company knowledge a GTM Brain. It includes things as small as your tone of voice. Ours says to explain complicated things in plain language, as you would to a primary school class.
Process. Write down how the process runs by hand today and where the bottlenecks are. Then note why automating it will save money or make money. We always start from the manual process. Our sales process automation guide covers what to validate before you build anything.
Two foundations come first
WorkflowsWho should build AI sales workflows
Salespeople should not build the automations themselves. Good ideas are usually unverified ideas, and people often start automating before they know whether the process makes sense.
Building it well takes knowing the tools, the data sources, how to join them, and how the data is structured. Then you pick the orchestrator: Zapier, Make, or n8n for plain automation, or Claude, ChatGPT, Codex, or Claude Code for AI orchestration. Every hour a salesperson spends on that is an hour not spent closing deals. Creating an agent from connectors and skills that are already set up, as in workflow 4, is the smaller job. Designing the data, processes, and orchestration underneath is the bigger one.
The salesperson's job is to run the process and give sharp feedback to the people who build it. The salesperson is the one executing it. A good request sounds like this: "This notification doesn't say it's meant for me." Or: "When a lead comes in, I need a context note on them to prepare for the meeting faster." Salespeople should take part in the tests and say what works and what doesn't.
The builder is a GTM engineering role. It combines two skills: understanding why sales and marketing work, and knowing how to use technology for them.
What stays human when you use AI in B2B sales
Anything that goes to a client or to the market should be checked by a person. Internally we accept a rougher draft, since the goal is passing information along quickly. Externally, ownership of what you say stays with you, whatever the model produced.
At Vanderbuild, nothing launches until we have read, fine-tuned, and corrected it by hand three times. Some steps later run on their own, such as the meeting email in workflow 1. For those, the check comes before automation: a person controls the workflow before agents take it over fully. It takes longer, and we accept that.
Relationships stay human too, especially in B2B. We build the trust and the credibility with clients ourselves and do not put agents in front of them. AI works well when it has clear guidelines, good context, and step-by-step instructions. A person stays accountable for the result. Our post Will AI Replace Sales Jobs? goes deeper into which tasks have left the salesperson, and which never will.
Generative AI for sales: what changed between 2023 and 2026
Three things changed in how we sell.
Where the work happens. One of us now spends 50% of their time in Claude or Claude Code. It is the operator we build systems in and deliver work from.
Time to value. A campaign used to take 1 to 3 days. Today we can prepare one in about 2 hours. Time to offer dropped from several days to 24 hours at most. That speed needs the process clear in your head first.
How the team is organized. Lead triage and information flow inside the company changed a lot. Marketing and sales both need to know who comes to us as a client. So we no longer split them, and they work as one growth team on one funnel. Go-to-market covers the top of that funnel, sales the middle, customer success the bottom.
One person can now cover a process end to end, where it used to pass from a marketer to pre-sales to sales. That is faster for the individual, and heavier, because one person carries more context at once.
What generative AI did not change is what makes a message work. AI tends to explain everything and use too many words. A good outbound message stays simple, short, and clear, and it lands when the reader has the problem right now. The tools changed how fast we work, and the work itself still starts with a clear offer and a written process.
Want AI workflows like these on your own pipeline? Book a free consultation and we will look at which one to start with.
FAQ
What is AI in sales?
Using AI to take repeatable work off salespeople: qualifying leads, drafting offers, answering replies, preparing campaigns. A person still owns what goes out and the relationship.
How can I use AI in sales?
Write down one manual process, where it stalls, and what it costs. Make sure the knowledge it needs is in one place. Then automate that one process and keep a person in control of it.
What tasks in sales can be automated with AI?
At Vanderbuild and at our clients, AI handles inbound lead qualification, CRM entry and enrichment, and offer drafts from call transcripts. It also handles replies on outbound campaigns, campaign preparation, and contract fine-tuning.
What are the top benefits of using AI in sales?
Speed and focus. At Vanderbuild, AI saves at least 10 minutes on every inbound lead and gets offers out within 24 hours. Salespeople can spend the time they get back talking to clients.
What is generative AI for sales?
The same models that write and summarize text, applied to sales work such as offer drafts, replies, and campaign copy. Keep the output short and have a person read it before it goes out.
.avif)