AWS PrivateLink#
AWS PrivateLink connects applications in an AWS virtual private cloud to supported services through private endpoints. Before creating resources for Point11, speak with your account manager and involve your AWS or network administrator. Explain what you want to connect, which accounts and regions are involved, and your access requirements. The AWS PrivateLink guide covers the AWS side. PrivateLink handles network connectivity; it is separate from the cloud account used for AI inference.
CRM data#
Bring customer data from a CSV export, HubSpot, or Salesforce: open Customers > New customers and choose CRM upload. Audience imports shape anonymous twins for a business profile; CRM Sync keeps named twins of real people in your CRM. CRM connections use read access, with one active connection per provider for your organization. Connect or disconnect them in Settings > Connections. In Customers > CRM, History in the menu beside an import or synced list shows its connection's Activity. Alpha connections are separate from these.
Audience imports#
In CRM upload, drop a CSV or Excel export, or choose HubSpot or Salesforce. A file opens Describe your customers with the file attached to your first message: Point11's customer sheet agent reads the whole file, decides where each column goes, and fills one row per customer, leaving names, email addresses, phone numbers, and other personal details off the sheet and saying which columns it left out. The file is kept with that Alpha conversation and deleted with it, and the agent's AI model providers keep none of it (zero data retention). A file holds up to 5,000 customers and 2 MB; split a larger one. Nothing is added until you choose Add customers. A HubSpot or Salesforce import opens on a table of your CRM's contacts, or Salesforce leads, which loads the whole list or record type behind a progress bar, then lets you search, filter, and sort all of it; filters offer only fields that hold values. Salesforce's All leads leaves out converted leads, which are already the contacts they became. Select the records to add, match their fields, and add them as customers; each field shows up to five sample values beside its suggested match. Fields that hold personal details, such as names, email addresses, phone numbers, street addresses, and birthdates, are never offered for matching, never read, and never appear in the table. An import is read on Point11's servers through your connection's read access, and it reads only the fields you match, importing their values as your CRM holds them. Fields that describe the account but match no field, such as its ecommerce platform or estimated revenue, import as Account context: Point11 keeps each one's name and value word for word, up to 2,000 characters, until you delete the import, and uses them to write customers based on that account. Your own grading or routing of an account, such as a tier, score, or account owner (your sales rep), isn't imported unless you match it. Public figures who own or front an account, such as a celebrity founder or the artist behind a brand, are kept as account context. Rows stay anonymous: imports shape your audience without copying individual contacts, and Point11 never links a row to a person in your CRM. Point11 stores the accepted values until the import builds your audience or you delete it; a failed import keeps them for a retry until seven days after it last failed. See Digital twins for details.
After your first one-off import, Imports appears under Customers > CRM and lists your uploads and imports with their status and the reason for any failure; a new upload or import opens it. From the list you can create customers from an import, retry a failed import, replace a CSV import's file or import from the CRM again, review its mapping versions, or delete it.
To keep a HubSpot or Salesforce list current, tick Keep updated beside the import and choose Monthly or Weekly; it's off unless you tick it, and a finished one-off import can be switched on from its row. Point11 imports the list again on that schedule with its saved mapping. Only a whole list or record type can stay updated: to keep a filtered group updated, save it as a list in HubSpot or a list view in Salesforce and import that; an import narrowed by search, filters, or picked records runs once. Once it's in, the list moves to Sync, where you change its schedule, update it now, or stop it; a stopped list returns to Imports.
Saved mappings#
Point11 saves each field mapping as a numbered version and reuses it when you import from the same source again. Point11 suggests the matches, and you can change any of them; columns recognized as personal identifiers, regulated data, or protected characteristics stay locked out. Each version records who decided each field. When your CRM adds, removes, renames, or retypes a field, Point11 flags the import's mapping for review in Customers > CRM > Imports instead of remapping it on its own, and imports keep using the saved version until you review it.
CRM Sync#
Follow a HubSpot contact, or a Salesforce contact or lead, to keep a named twin of that person. Once a CRM is connected, Sync appears under Customers > CRM. Choose Follow contacts to load the CRM's contacts and search, filter, and sort them; for Salesforce, Type switches between contacts and leads. Select everyone in a HubSpot list, a Salesforce list view, or the whole record type to follow it as a segment that stays in sync, and turn on auto-follow so new members are followed as they appear; otherwise pick up to 100 contacts at a time. Records already followed, forgotten, merged, or converted are skipped, and Point11 says how many. Follow settings in the same dialog set where history starts, which HubSpot properties hold transcripts, and sharing with your organization. Sync has two parts: Live syncs lists every kept list and followed segment with its schedule and controls, and Followed contacts lists each contact kept as a named twin with its status. Point11 reads the contact's name, the fields your contacts mapping selects (such as title, department, and account), and the call and meeting transcripts linked to it, and builds the twin from those records. Salesforce call transcripts need Einstein Conversation Insights.
Changes sync on their own. Point11 checks followed HubSpot contacts for changes every hour and Salesforce records every few minutes. Point11 waits a few minutes after a change so several edits become one update, and updates a twin only when its mapped fields or linked calls change. Each update adds a version to the twin's history that shows what changed, and Update now requests one at any time. If you edit a field on a synced twin, later CRM updates keep your value; when the CRM changes that field, Point11 asks whether to keep yours or use the CRM's. Each campaign's next run uses the twin's newest version, and finished results keep the version they ran with.
Your organization can follow up to 1,000 contacts. Contact updates are included in your plan; when your organization reaches its daily included capacity, remaining updates wait and run automatically later. If a connection loses access, updates pause until someone reconnects it. Pause a contact to stop its updates. A contact that leaves a followed list stays followed until you forget it.
Followed contacts are personal data. Everyone in your organization can see them. Forgetting a contact in Point11, or deleting it in the CRM, stops these reads at once and erases its personal data after 30 days; until then a member can restore it. When the CRM merges a followed contact into another record, or converts a followed Salesforce lead into a contact, Point11 retires the old record the same way and follows the surviving record while your organization is under its follow limit. Billing records and finished campaign results are kept.
AI inference#
Point11 provides AI model access by default, which Settings > AI inference shows as Vercel AI Gateway under Provider. To use your own cloud account, choose your cloud under Provider, choose your own account under Billed to (such as Your AWS Bedrock account), enter its credentials, and choose Save. Your cloud provider bills supported calls made with that key.
- AWS Bedrock
Enter your Access key ID, Secret access key, and Region.
- Azure AI Foundry
Enter your Resource name and API key.
- Google Vertex AI
Enter your Project ID, Location, Service account email, and Private key.
We make a test call before saving. A saved key shows Verified and the date, or Test failed. You can Re-test it (Key verified confirms it worked), Replace it, or switch Billed to back to Point11's account in the same place. Saved secrets are encrypted and aren't shown again.
Your choice applies to compatible calls across Point11 and Alpha; the selected cloud must support the model being used. Under Usage, each key shows Calls on your account and their Input tokens and Output tokens, apart from calls that Ran on Point11's account and those Not matched yet; compare them with your cloud bill.
Export your results#
Export Discovery reports as CSV or NDJSON, copy Alpha documents as Markdown, or download presentations as PowerPoint files. Admins can export the audit log. Your own software or assistant can also read campaign information and request Alpha reports through the API or MCP server.