What is CRM Agent and how is it different from a regular CRM?
In a regular CRM a person fills in the forms: the same person then searches for repeats, moves deals through stages and prepares invoices. Here it is the other way round — an AI agent keeps the data: it creates contacts and companies, cleans the base, merges duplicates, accepts requests from the site, moves deals through the pipeline, prepares quotes and invoices, assembles mailings and serves requests. You watch through the screens and give commands in words. That is exactly why the interface does not have a single data-editing form — it is both protection from accidental edits and the very idea of the product. Details in Getting Started and Screens & Workspaces.
Is an Xedant Agent required?
No, the product works without it. Without the agent everything is available except the chat with the AI agent: the client base, the timeline, data quality, import and exports, custom fields, leads, deals, money, marketing, communications, service, screens and management over the API. Just keep in mind: the agent is exactly what does all the routine work, so without it the data has to be changed by hand over the API — and that is programmer’s work. Details in Chat with the AI Agent.
Where is the data stored?
Entirely on your server. By default the embedded database in the product’s data folder is used; your own PostgreSQL can be connected if you like. Neither an account with us nor someone else’s cloud is needed. Communication with the AI model goes only through your Xedant Agent — that is, to wherever you yourself pointed your agent. Details in Installation.
What is included in the product and what are add-ons?
The product is assembled from a core and add-ons. The core holds the minimal foundation: the client base, the timeline, data quality, import and exports, custom fields, management over the API and the screens. Everything else is switchable add-ons: Deals, Money, Leads, Marketing, plus the communications and service add-ons. A fresh installation has only the core available, and the contours you need are switched on by choice in the Addons catalogue. Switching on and off applies on restart — that is more reliable than changing the composition on the fly. Switching off keeps the data: the code turns off, while the records stay in place. Details in Add-ons & Product Composition.
How do I get the data out of an old CRM?
By importing from a CSV file: first comes a trial run without writing — you see what will come out — then the real run, and if something went wrong, the whole run rolls back. There is linked import: contacts are created together with their companies, for example. Outward, data is exported as any set with its history, and bulk edits and deletions can be undone too. All operations — import, export, a bulk run, a merge and mail synchronization — are visible in the single history. Details in Import, Export & Bulk Operations.
How does it fight duplicates?
The product searches for them itself — on every creation and change of a record, by email, phone, company details and domain. It shows an exact and a fuzzy match score and the trait the records matched by, and the review happens in a separate queue: Pending, Not duplicates, merged. Then — comparing the participants side by side, merging with undo, and bringing the data to one form with AI (also with undo: phones, names, company names). Separately there is verification and enrichment of the counterparty’s data with the source and its age noted, so it is always visible what is verified and what is assumed. Details in Data Quality & Duplicates.
Can requests be accepted from a site?
Yes. There are public intake addresses with a key — they can be put into a form on your or someone else’s site — and a ready public form page. All channels work by one intake contract: a request turns into a lead the same way, and repeat approaches are glued together before assignment. A lead always gets an owner — not one is left without a master — and if no answer comes in time, the first-touch deadline control and escalation kick in. Nothing is lost: rejected requests and spam are stored too. Details in Leads.
Do email and calls work?
Email is synchronized over IMAP (the mail protocol letters are fetched from a mailbox by): the correspondence lands in the client’s timeline by itself, and email chains are assembled into threads. A request from any channel — an email, a call, a messenger, the site widget — becomes a conversation in the shared inbox, with assignment, statuses and deadline counting from the moment it arrived. Honest borders: telephony is a junction with an external service, and call recordings and transcripts require a human check; messengers are configured as data and need your own connection at the provider. Details in Communications.
Is there support for personal-data requirements?
Yes. There are retention periods and deletion of personal data on request, a server-side split of internal and public messages (an internal note never goes to the client), and conversation privacy on all screens. The data sits on your server, and only what you sent through your agent goes to the AI model. Details in Data Quality & Duplicates and Service.
How are access rights arranged?
There are two different accesses. The browser you watch the screens in can only read: no form can change the data. All changes are possible only with the management key you give to your agent. Separately there is a limited key for request intake: it can only create a lead and never touches the client base. Every edit shows its actor — Agent or a human. Details in Management API.
What is needed besides the product itself?
A server or a computer with Docker — nothing else is mandatory. The database is embedded, and a separate one is needed only at large volumes. The AI model connects through your Xedant Agent, so if the chat with the agent is not needed, a single container is enough. An Xedant Agent license is already included in the product license. Details in Installation.
How much does it cost and what does the license restrict?
The trial period is 30 days free. After that the prices are the same as the neighboring products:
Did not find the answer — take a look at the documentation or start with the installation.