AI sales agent: account research before every call

Profile

A good sales conversation starts with knowing who you are talking to: what the business does, what changed recently and what you have already done together.

Finding that out takes time an account manager does not have between two meetings, so it often only half happens.

The Prospect Researcher prepares a briefing before every meeting from the company's website, the Business Register, recent news and the history in your CRM. Getting in touch is something the account manager does themselves.

What is an AI sales agent, and which part does this one do?

An AI sales agent is an AI agent that carries out sales tasks in your own systems, such as your CRM, your mailbox and your calendar. It reads what comes in, looks up information, writes drafts and updates the CRM. This AI employee does one part of that: it researches an account before the conversation and prepares a briefing.

The difference with a chatbot on your website is big. A chatbot talks to visitors. This agent works behind the scenes, for your own team. So it is more like an inside sales assistant who never gets tired than a salesperson.

The difference with the automation your CRM already has is worth knowing too. A workflow sends a fixed reminder after three days. An agent first reads what the lead wrote, checks whether there is already a meeting booked and only then decides what should happen. Sometimes that is an email. Sometimes it is nothing at all. What an agent is exactly, you can read in what is an AI agent?.

We use this ourselves. When an enquiry comes in through our own site, an agent looks up what is publicly known about the company: what they do, for whom, whether they are hiring and which systems they mention. That briefing then sits as a note on the deal in Pipedrive. It only mentions what it has actually read, and what it does not know is listed too. That last part is often the most useful.

What it does, step by step

Spot the meetingA new appointment in the calendar, or an account you name yourself.
Research the companyWebsite, KVK details, recent news and public announcements.
Pull in the CRM historyEarlier contacts, quotes, orders and open questions.
Write the briefingWhat the company does, what is going on, what has already happened and which questions are worth asking.
Have it ready for the meetingIn the CRM and on the calendar appointment, with the sources attached.

Which sales tasks does an AI sales agent take over?

An AI sales agent takes over the tasks that keep coming back and need little judgement: researching accounts, following up leads, preparing and following up quotes, keeping the CRM tidy and picking up signals at existing customers. We have worked those tasks out as separate AI employees. That way you start with one task and only expand once it runs well.

Sales task	What the agent does	AI employee
Account research	Summarises the company, news and CRM history in a briefing	This page
Following up leads	Responds to new leads, asks qualifying questions and routes the right ones to sales	Lead Responder
Preparing a quote	Proposes a quote based on earlier proposals, agreed prices and the request	Quote Assistant
Following up a quote	Keeps track of open quotes and flags what needs attention	Quote Follow-up Assistant
Keeping the CRM tidy	Finds duplicate records, incomplete details and outdated contacts and suggests corrections	CRM Cleaner
Signals at customers	Follows what happens at existing customers and flags when a conversation makes sense	Account Signals Assistant

The agent mainly saves time on work that nobody misses once it is gone. Often the gain starts before the agent, with an enquiry that lands in the CRM correctly straight away. For an EV charger supplier we built a configurator that puts every enquiry straight into the CRM, with the full configuration attached, so nobody has to retype anything.

Onboarded into your business

Niels does not start from a generic script. Like any new colleague, Niels is onboarded into the way your team works first.

Your knowledgeWe onboard Niels with your own documents, examples and agreements, so the work is done the way your team does it.
Your systemsWe set up the integrations for the software you already use. Different fields or different steps? We adapt to them.
Your rulesYou decide what Niels handles on its own and what always goes to a person first.
Grows with youWhen your way of working, a system or a rule changes, we adjust Niels to match.

Works with these systems

CRM

Sources

KVK
APIs

Calendar

Communication

Is your system not listed? We can connect any system, even if there is no standard integration. View all integrations.

Does an AI sales agent work in HubSpot or Pipedrive?

Yes. The agent works through the API of your CRM. It reads contacts, deals and notes, and writes tasks, briefings and drafts back. In HubSpot and Pipedrive that is possible with a custom integration or with the package's own AI features. Which route fits depends on your process and your other systems.

If you use HubSpot, there is already an agent built into the package. In Sales Hub, HubSpot offers a Prospecting Agent that tracks buyer signals, finds contacts and drafts personal messages (HubSpot, accessed October 2026). According to the same page, you can review drafts first, or let the agent send emails in an autonomous mode. I advise against that second option, for the reason above.

If you work with Pipedrive, or with a mix of systems, we often build an agent of our own that works through the API. What is technically possible is described under the HubSpot integration and Pipedrive integration.

An agent is only as good as the data it looks at, however. If your CRM is full of duplicate contacts and empty fields, an agent just makes that mess faster. In that case, start by cleaning up. It is less exciting, but it pays off straight away. One warning: merging records cannot be undone in HubSpot, Salesforce or Pipedrive. That is why we only do it after approval, in batches that can be rolled back.

Better still is stopping the mess before it starts. For an installation company we built an integration that puts enquiries into the CRM and spots duplicates by the same email address, postcode and house number. If the same visitor comes back shortly afterwards, the integration adds to the existing enquiry instead of creating a second one.

What runs automatically and what does not

You stay in control.

Not every decision should be made automatically. For each workflow, we set out:

What can happen automatically.
What is put forward for approval first.
When a person is brought in.
What information is stored.
How actions are logged.

What stays with the salesperson?

The relationship, the negotiation and the pricing decision stay with the salesperson. The agent can prepare a conversation and draft a quote, but it does not decide on discounts, terms or an exception for a loyal customer. That takes knowledge of the customer, a feel for timing and responsibility. You do not leave those three to software.

After all, a customer buys from someone they trust. That person calls when something goes wrong and knows the buyer's name. A good negotiation also revolves around what is not in the email: the customer has doubts, the competitor is cheaper or the budget holder has just been replaced. A salesperson picks up such signals in a conversation, not in a CRM field.

The rule I use: anything a customer gets to see goes past a person first. Only when a task has run well for months do we discuss whether part of it can go without a check.

What I would not do: put an agent on a list of two thousand prospects and let it send emails on its own. One odd email with the wrong name gets forwarded and sticks. So let the agent research and write, and let the salesperson press send.

What I tell a salesperson who fears for their job: the agent takes over the research, not the customer. You keep more time for conversations, and that is where the selling happens.

Is this a fit for your organisation?

A good fit if

Account managers with many new meetings a week.
Sales teams selling to businesses, with a longer sales cycle.
Organisations with a CRM full of history that nobody reads before a meeting.

Less of a fit if

Contacting prospects itself: cold emails and messages fall under the GDPR and the rules on unsolicited communication, and we leave that to people.
Transactional sales with many small customers, where a briefing per meeting is too much.

What it can deliver

Meetings that start with knowledge of the customer.
Less searching between two meetings.
CRM history that gets used instead of forgotten.

Directions, not promises. As soon as there is evidence from a client project, it will appear here as a substantiated result.

Frequently asked questions

Is this an AI sales agent that contacts prospects itself?

No. The terms AI sales agent and AI SDR are often used for software that sends emails and messages on its own. This AI employee does the research beforehand and prepares a briefing. Getting in touch stays with your account managers, partly because unsolicited messages fall under the GDPR and the rules on unsolicited communication.

Which sources does it use for an account briefing?

The company's website, the Dutch Business Register (KVK), recent news and public announcements, and the history in your own CRM. Every statement in the briefing comes with its source, so the account manager can check where it came from.

Can you use AI to gather information about prospects?

Data about a company is usually not personal data; for a sole trader that is different. That is why we limit the briefing to the company and the contacts already in your CRM. What is recorded and for how long is agreed in advance.

How is this different from the Account Signals Assistant?

The Account Signals Assistant follows your existing customers continuously and flags when something is going on. The Prospect Researcher works per meeting: before an appointment, it sets out everything you need to know about that company.

Where does the briefing end up?

Where your account manager looks for it: on the account in your CRM, on the calendar appointment or as a message in Teams. Nobody has to open an extra tool before the meeting.

What does an AI sales agent cost?

That depends on the task, the number of systems and whether you use a feature of your CRM or have an agent of your own built. Which factors drive the price, you can read under AI agent development costs.

Will an AI sales agent replace my salespeople?

No. An AI sales agent takes over the work around the sale, such as account research, follow-up and keeping the CRM up to date. The relationship, the negotiation and the pricing decision stay with the salesperson.

Which task should you start with when you bring in an AI sales agent?

With one task that comes back often and needs little judgement, such as account research before a meeting. Once that runs well, you expand to follow-up or keeping the CRM tidy. Which task pays off most for you, we work out in the AI Opportunity Audit.

How to put this to work

1A 30-minute callTogether we look at your process and your systems, and whether this task is the best place to start.
2Set up on your rulesIntegrations with your own systems, and agreeing what may run automatically and what a person approves.
3Measure and adjustFirst alongside your team, only then on its own. We look at what it delivers and adjust.

What does an AI employee cost?

More sales tasks

See all AI employees for sales