Today
The same customer sits in the CRM three times, spelt three ways, with quotes spread across all three. Many contacts are out of date: moved on, retired or promoted, and the mailing comes back as bounces. Nobody wants to clean up because every check takes time. So sales stops trusting the CRM and keeps its own lists.
An example
Orvelon has 14,200 contacts and 3,100 companies in its CRM. The agent finds 410 duplicate companies, among them a dairy entered with three spellings whose quotes for the cleaner AR 240 are spread across all three. From bounces and out-of-office replies it identifies 380 people who have left their companies, and from signatures 95 new contacts, such as the dairy's new production manager, who has been copied on e-mails for three months but is missing from the CRM. The CRM owner approves the proposals within two days, in batches and with evidence for every change.
Your data, wherever it lives
Who does what at your customers is in the CRM today, but the current state sits elsewhere: in signatures and out-of-office replies in the mailbox, in bounces after the last mailing, in meetings with new faces and in the legal notice on the customer's website. demi reads, understands and links it per company and person, with your rights. The cleanup nobody has time for is ready as a list of proposals within hours, and you decide what is merged, changed or deleted.
Where the data lives today
- CRM
- E-mails
- Web & portals
- Calendars
How demi solves it
Three layers in this order: connect the data first, then the rights, then the AI.
- 1 · ConnectData Layer
Data sources in this case: the CRM, e-mail, calendar and your customers' websites. demi replaces nothing and matches companies and contacts in the CRM against signatures, distribution lists, out-of-office replies and bounces, against the attendees of your meetings and against the legal notice and contact pages of customer websites.
- 2 · PermitPermission Layer
For the cleanup the agent uses only senders, recipients, signatures, out-of-office replies and bounces from the mailboxes; e-mail contents stay with those who may see them. Each sales person sees the proposals for their own customers. Whoever owns the CRM approves merges and deletions, and every change is recorded with its evidence.
- 3 · ActIntelligence Layer
The agent spots duplicates by name, address, e-mail domain, VAT ID and phone number, even with different spellings, and proposes which record stays. It derives job changes from bounces, out-of-office replies and new signatures, fills gaps such as role or direct line with their source and flags contacts nobody has spoken to in years.
How it runs
- 1
demi reads the CRM and matches companies and contacts against e-mail traffic, meetings and customer websites.
- 2
The agent finds duplicates and proposes which record stays and what is merged.
- 3
From bounces, out-of-office replies, signatures and distribution lists it recognizes who has left, changed role or is new.
- 4
It fills gaps such as role, direct line or industry with their source and flags old contacts under your deletion rules.
- 5
Whoever owns the CRM approves the proposals in batches; merged records keep their history.
What changes
One record per customer, with all quotes and contacts
Contacts who have left are caught before the next mailing
A CRM that sales trusts again
What does it bring you?
Check it with your own numbers
Every starting value is an assumption you can change. The calculation sits right below, step by step.
Your numbers
Result
+96 hfreed per year
Assumptions and calculation
demi finds duplicates, job changes and gaps with evidence; you check and approve the cleanup.
6 min → 3.6 min per case
How we calculate
200 contact records × 6 min × 40% × 1.0 (manual) × 12 months = 5,760 min = 96 h
Value per year, before demi costs
Freed per year+96 h
Assumption: labor cost of $48.62 per hour in US manufacturing in June 2026 (BLS), plus workplace, IT and overhead.
= Value per year, before demi costs
+$6,240
gained per year
96 h ÷ 1,760 h per role = 0.1 full-time roles
An estimate, not a promise. The starting values are assumptions for an industrial company with 150 to 500 people.
Combine with more cases and get the packageQuestions about this use case
Does demi delete contacts by itself?
No. The agent proposes; the responsible person approves. When records are merged, activities, quotes and e-mails move to the remaining record, so nothing is lost. Deletion happens only after approval and under the deletion rules your company has set.
Does demi read all employees' e-mails for this?
No. The agent uses only mailboxes released for this purpose, and from them only senders, recipients, signatures, out-of-office replies and bounces. What the e-mails say remains visible only to those who could see it before.
How does the agent tell that two companies are the same?
By name, address, e-mail domain, VAT ID and phone number, even with typos and abbreviations. Each proposal shows which features match. If two records belong to one group, such as a plant and its head office, it proposes a link instead of a merge.
Will the CRM stay clean after the cleanup?
demi keeps checking. New duplicates, job changes and gaps go once a week as a short list to the person who owns the CRM. Approval works as in the first round, only with far fewer cases.
