And Dave is also dealing with three customers, answering email and trying to remember who asked for a quote on Tuesday.
That's not a criticism of small businesses.
It's how small businesses work.
People cover multiple functions because there aren't enough people to create a department for every piece of work.
Agentic AI becomes interesting because some of those missing capabilities can potentially be provided without adding another full-time role for every function.
The right question
It's not about replacing your salesperson.
For a small business, this distinction matters enormously.
If you have one salesperson, the sensible starting question probably isn't:
How do we replace them with AI?
It's:
What are they doing that doesn't actually need a salesperson?
Researching a company for ten minutes before a call.
Copying information into the CRM.
Writing up notes.
Preparing the same type of proposal again.
Checking whether somebody replied.
Remembering to follow up.
Finding product information.
Working out which enquiries need attention.
Updating opportunity stages.
Creating tasks.
Those things matter.
But they aren't necessarily where your salesperson creates the most value.
LET PEOPLE DO THE SELLING. Let technology carry more of the machinery around it.
Where it could help
Where could a small business use an AI sales agent?
You don't need a giant transformation programme.
Look at the ordinary work.
01
Website enquiries
Make sure genuine enquiries don't sit waiting.
Someone completes your form at 6.45pm.
Perhaps nobody sees it until tomorrow.
An agent could potentially:
read the enquiry,
understand what they need,
check relevant business information,
identify missing details,
prepare a response,
create or update the CRM record,
and make sure there is a next action.
Depending on the boundaries you set, a person might approve the response before anything is sent.
THE WIN · Faster response. Less administration. Fewer forgotten enquiries.
02
Lead qualification
Find out what you need before somebody spends time on it.
Not every enquiry is ready for a sales conversation.
An agent could help gather:
what the person needs,
location,
timescale,
budget where relevant,
quantity,
requirements,
and other information your business genuinely uses to qualify opportunities.
Then it can organise that information for a person.
THE WIN · Your team spends more time on conversations that are ready for them.
03
Follow-up
Because "I'll remember to chase that" isn't a system.
This may be one of the least glamorous and most useful places to start.
An agent can potentially keep track of:
quotes awaiting response,
people who asked you to come back later,
actions agreed in meetings,
enquiries waiting for information,
and opportunities with no next step.
It could prepare the appropriate follow-up and either ask for approval or act within clearly defined limits.
THE WIN · Fewer opportunities disappear because everyone got busy.
04
Meeting preparation
Walk into the conversation knowing what's happened.
Before a meeting, an agent could bring together:
who the customer is,
what they've previously asked,
emails and conversations,
what they bought before,
what they're interested in,
open questions,
promises already made,
and relevant company information.
THE WIN · Less preparation. Better context.
05
Quotes and proposals
Stop starting from an empty document.
If your proposals repeatedly draw on the same:
service information,
product information,
pricing,
terms,
customer requirements,
case studies,
and company information,
an agent could potentially prepare the first version.
A person can then handle the commercial judgement and final approval.
THE WIN · Less document assembly. More time on the part that actually needs expertise.
06
CRM administration
Your CRM can't help much if nobody updates it.
An agent could potentially:
log interactions,
summarise conversations,
update agreed fields,
create tasks,
record next actions,
and identify missing information.
THE WIN · Better records without making your salesperson become a data-entry clerk.
07
Prospect research
Do the gathering before somebody needs it.
An agent could research a company, combine that with information already in your systems and prepare a short brief.
Not 27 pages of AI-generated research.
The few things somebody genuinely needs before making contact.
THE WIN · Useful context without the tab marathon.
08
Pipeline attention
Tell me what needs me today.
For a small business owner, this could be particularly useful.
Instead of opening the CRM and trying to reconstruct what's happening, imagine starting with:
TODAY
3 new enquiries need review
2 quotes are due follow-up
1 customer replied overnight
1 opportunity has no next action
2 routine follow-ups are prepared
Nothing else needs you
That may be more useful than another dashboard.
THE WIN · Attention goes where it's actually needed.
The first big win
The first big win may simply be remembering.
There's a tendency to imagine AI agents doing spectacular things.
Negotiating deals.
Running autonomous outbound campaigns.
Managing whole pipelines.
For many small businesses, the first useful agent may do something much more ordinary.
IT MAKES SURE THINGS HAPPEN.
Every genuine enquiry gets seen.
Every agreed action gets recorded.
Every quote gets the appropriate follow-up.
Every customer question gets routed.
Every important exception reaches a person.
That can matter because small-business processes often rely heavily on individual memory.
AI doesn't need to become your salesperson to improve that.
No dedicated sales team?
What if you don't have a sales team?
Then this may be even more relevant.
Many small businesses don't have somebody whose only job is sales.
Sales might be handled by:
the founder,an account manager,an administrator,a technical specialist,or whoever is available.
Which means every hour spent on sales administration comes from somewhere else.
An agentic workflow can potentially give that person support around the sales process without pretending the business suddenly has an autonomous sales department.
Referral-led sales
What if most of your sales come through referrals?
Agentic selling can still be relevant.
This isn't synonymous with cold prospecting.
Imagine a referral arrives.
You still need to:
respond,
understand what they need,
collect information,
prepare for the conversation,
send something afterwards,
prepare a quote,
record what happened,
and follow up.
The lead source changes.
The work around the lead doesn't disappear.
A few leads only?
What if you only receive a few leads?
Then you may not need an AI sales agent.
That's important.
If you receive five enquiries a month and handling them takes very little time, building an agentic workflow may solve a problem you don't have.
Ordinary software or a simple automation might be enough.
We look for repeated work with enough value or friction to justify changing it.
AI should earn its place.
Your existing tools
You don't need an enterprise technology stack.
Another misconception is that agentic AI requires a giant collection of systems.
It doesn't necessarily.
A small business might already have:
a website,email,calendar,CRM,documents,accounting software,and perhaps an automation platform.
The question is whether those systems can provide the information and actions required for the workflow.
Sometimes an agentic layer can work around what already exists.
Sometimes an integration needs building.
Sometimes the existing process needs tidying first.
But:
DON'T BUY SIX NEW AI TOOLS BEFORE YOU'VE IDENTIFIED THE JOB.
