AI in sales
ABM and AI agents: how to track changes at target accounts every week
In brief: An AI agent helps you monitor target accounts when you give it a defined set of companies, a weekly task and clear evaluation rules. These companies can be existing key accounts or a limited group of prospective customers. The agent compiles changes it finds in personnel and responsibilities and prepares draft messages, but a salesperson checks the interpretation and approves any outreach.
Connect NordicContacts to your AI assistant and start monitoring your target accounts.
How do you keep up with changes at your chosen target accounts?
Your target accounts are already selected, sales responsibilities assigned and key conversations recorded. Yet a new CFO can still go unnoticed, or you may only discover a sales director's expanded responsibilities at your next meeting. The challenge is not finding companies but keeping track of what is happening at your target accounts.
Weekly monitoring makes this a repeatable task. The agent reviews the defined group of companies, retrieves changes for the selected period and highlights findings that could affect an account plan or your approach to a prospective customer. The salesperson does not need to review each company individually.
We will use a hypothetical example throughout this article: you sell production planning software to Finnish manufacturing companies. At your target accounts, you are interested in the people responsible for production, finance and sales. A new CFO may change how investments are assessed, but a name appearing in search results does not yet tell you anything about that person's objectives.
At an existing key account, a finding may prompt a review of the account plan. At a prospective customer, the same finding helps you assess whether there is a timely reason to get in touch. In both cases, monitoring should give the salesperson a finding to evaluate, not decide on outreach on their behalf.
How do you limit monitoring to a customer list or segment?
Start with companies, not people. You can define the target accounts to monitor in two ways: a customer list you upload yourself or a segment defined by search criteria. The choice depends on whether you want to monitor a named group or also include new companies that begin to meet your criteria.
| Approach | How is the group of companies defined? | What is it suitable for? |
|---|---|---|
| Customer list | You upload a list of named companies. The list contains companies only. | Monitoring existing key accounts. |
| Segment defined by search criteria | You specify criteria such as industry, company size and location. New companies that meet the criteria are also included in monitoring. | Monitoring prospective customers. |
You can restrict an MCP search to companies on a customer list. Give the agent the customer list to monitor and instruct it to retrieve changes only from those companies. This keeps the weekly task focused on the same approved group rather than letting it drift into other companies in the same industry. You can also exclude companies on a customer list from a search. When monitoring a segment of prospective customers, exclude your existing customer list. Make this choice before searching for people so the agent does not prepare a prospecting message for an existing customer.
In our example, the manufacturing companies are monitored as a segment defined by search criteria, not as an uploaded customer list. The criteria stay the same from week to week, but the companies matching them may change. The scope approved by the salesperson therefore guides monitoring, even though not every company has been named in advance.
How much is there to monitor in the example target group?
Our concrete example segment consists of manufacturing companies in Finland with a known employee count of 20–249. It includes only companies with at least one contact who has a work email address, a phone number or both. This scope excludes other employee ranges and companies whose employee count is unknown.
In the NordicContacts database, 1,262 companies matched these criteria on 20 October 2026. The change figures for the same example group, confirmed in October 2026, illustrate the monitoring workload:
| Metric | Review period | Number of companies |
|---|---|---|
| Companies in the database matching the criteria | Total | 1,262 |
| Companies with new or updated decision-maker information | 7 days | 186 |
| Companies with new or updated decision-maker information | 30 days | 501 |
| Companies with new contacts | 7 days | 60 |
These figures count companies, not contacts or job changes. The groups may overlap, so they should not be added together. This is an example segment in the database, not Finland's entire market or evidence of any individual's decision-making authority. Results for your own monitored group will depend on its criteria and whether you exclude existing customers.
What should the agent check each week?
NordicContacts search includes filters for new contacts and new or updated responsibility information within a selected period, such as the last 7 days. The agent can therefore directly request last week's changes at companies in a customer list or segment. You do not need to save an earlier search result for comparison before starting weekly checks.
For our example, limit the task to people responsible for production, finance and sales. Ask the agent to group results by company and distinguish new contacts from new or updated responsibility information. This shows the salesperson which companies need a closer look and which roles the findings concern.
A new contact does not automatically mean a new employee: the record may have been supplemented or the job title refined. Updated responsibility information alone does not confirm a role change either. Ask the agent to explain precisely which information is new or updated and avoid concluding that “the CFO has changed” unless the actual change has been confirmed separately.
How do you set up weekly monitoring with MCP?
Connect NordicContacts to an AI assistant—Claude, ChatGPT or another MCP client—by enabling the MCP service. Create a scheduled weekly task in the assistant using the instructions below. The agent uses change filters to retrieve changes at companies in the segment or customer list, summarises its findings and prepares draft messages.
An effective brief for our example segment could be:
Every Monday, check companies in my defined manufacturing segment for new contacts and updated responsibility information from the last 7 days. Exclude companies on the existing customer list. Limit the check to people responsible for production, finance and sales. Group findings by company, distinguish new contacts from new or updated responsibility information, and explain the basis for each finding. Prepare a draft message for relevant changes. Do not send any messages.
Ask the agent to report any failed checks. If a search is incomplete, it must not present missing results as employee departures or a completely changed organisation.
Test the workflow on a small subset of your target accounts first. Compare the agent's summary with a manual review and refine any overly broad job-title criteria before applying it to the full group. This helps prevent roles such as production manager and production director from being treated as the same monitoring target without you noticing.
If you want to build monitoring into your own system, you can also use the REST API, and transfer data to your CRM through CRM integrations.
How do you turn a change into a useful draft message?
Alongside the change information, give the agent context about the account or sales conversation: what has already been discussed, which problem has been identified at the company and who owns the next contact. If there has not yet been a conversation, say so. Without this background, the draft can easily become a generic congratulatory message that does not help the recipient.
The software salesperson in our example can instruct the agent to distinguish between new finance responsibilities and production responsibilities. For a finance role, it prepares a question about evaluating investments; for a production role, a question about practical planning challenges.
| Finding | Message angle | Example question |
|---|---|---|
| The person responsible for finance is new in the monitoring results | Evaluating investments | “Would it be useful to discuss how the impact of a production planning improvement project could be assessed from your perspective?” |
| Production responsibilities have expanded | Planning across the area of responsibility | “With your expanded responsibilities, is there a need to review shared production planning practices across your area?” |
| Information about the person responsible for sales has been updated | Collaboration between sales and production | “Is aligning delivery commitments with the production plan a current topic of discussion at your company?” |
Show the finding, the account background used and the proposed recipient alongside the draft. The salesperson checks whether the change is real and the message timely. The message is sent only after approval.
How do you assess the value of monitoring and keep responsibilities clear?
Keep the weekly summary short and turn it into a clear work queue. Put companies where a change relates to an open opportunity or an agreed account plan first. Assign an owner to each proposed action and record a decision: make contact, investigate further or take no action.
Limit personal data and customer notes shared with the assistant to what is necessary for the task, and agree on access rights, retention periods and how to respect contact opt-outs.
Connect NordicContacts to your AI assistant and start monitoring your target accounts.