Territories, quotas and capacity. One plan, on live data.

No. Map street, city, state and postal fields in the recipe output and geocoding runs there, once, with results cached by normalized address. Rows that already carry latitude and longitude are used as they are and never re-geocoded.

No. Salesforce is the connector that ships today, spreadsheets need no connector at all, and other CRMs or systems your data lives in are configured per instance. The sync layer is source-agnostic: objects, filters and modification watermarks work the same way whatever the source.

Both, on the same plan. A quota plan attaches to an alignment and uses its territory tree and metrics. Allocate top-down with six methods, floors, caps and a period split; let managers propose bottom-up within a guardrail; cover territories with people, ramps and shares; compare capacity with quota; pull actuals for attainment; and sign off through the same approval flows. Adjustments, versions and the drift check keep the numbers right through the year. Quotas leave through a compensation export today; publishing them to CRM forecasting is on the way.

No. In suggest mode it reads and drafts only. In edit mode it creates proposal cards with the exact changes and their impact; nothing happens until a person clicks Apply, and applied changes are logged as made by the assistant on that person's behalf. It never publishes, syncs, deletes, restores snapshots, or touches users and settings, and it runs with exactly the permissions of the person asking.

Yes. Every plan and every alignment has a team chat, and any territory, record, boundary or quota plan can carry comment threads with mentions and replies. Mentions, replies and comments on territories you own arrive in the in-app inbox, next to the approvals waiting on you. Discussion stays open on approved scopes, so reviewers can talk about what is locked without changing it.

You choose Anthropic or OpenAI and provide your own API key; the model is picked per instance from a list with prices. The assistant runs inside your instance and sends the model only what the request needs. Optional data masking replaces names, contacts and record ids with placeholders before they leave. Token budgets cap usage per user per day and per instance per month, and every request is reported with an estimated cost.

Yes. A publish template maps territory names, owners or any attribute to standard or custom fields on the source object. Each publish job shows the destination impact before it runs, runs once, and keeps a per-row log. A plan can require an approved alignment before a job may run.

Yes. Approval templates define the steps and who signs each one. A submission can cover the whole alignment or selected territories, every step carries a due date, reviewers see the exact diff, and approved territories lock. Administrators can reverse an approval, in whole or for selected territories, and the reversal is on record.

In your own instance: a dedicated API and a dedicated PostGIS database that no other customer shares. Cloudflare sits in front for TLS and traffic protection. Connection credentials, Salesforce tokens included, are encrypted at rest with a key that belongs to the instance.

Sign-in is OpenID Connect with your identity provider, Okta, Entra ID, Google, Cloudflare Access, Auth0, Keycloak or Salesforce, or email and password. SCIM 2.0 creates, updates and deactivates users and maps groups to roles. API tokens act as a named user under that user's permissions, and schedules run syncs, pipelines, geocoding, actuals refreshes and publish jobs on a cadence, each as the person who created it and with the same locks.

Yes. Excel and CSV import needs no CRM connection: the wizard maps columns, declares types and picks an upload mode, and remembers everything for the next file. Connect a CRM later and the same plans keep working.

Eleven standard sets ship with every instance: US states, counties and ZIP areas, Canadian provinces and forward sortation areas, world countries, UN sub-regions and states and provinces worldwide, and Europe's NUTS 1 to 3 regions, 45,169 boundaries with filterable attributes such as country, region, city or population. Any other region is added from its official boundary data. Any ZIP or postal list of your own becomes a boundary set, and territories can be grown from their centers as radius or drive-time shapes with the parameters you set.

A plan of 300,000 records and 2,000 territories, with 2,000 sellers covering them and a quota plan allocated over every territory, has been exercised end to end. At that size the workspace opens in about twenty seconds, rule and optimizer previews answer in seconds, people, capacity and export views in well under a second, and a whole-tree quota allocation runs as a background job of about three minutes, measured on one development machine. No computation caps rows: metrics, health checks, quotas and publishing read every record, while the browser loads up to 100,000 records at a time and says so when there are more. Production instances are sized per customer; tell us your counts.

An instance provisions from a blueprint and restores the boundary reference data on its first boot. Connecting Salesforce is a standard authorization and the first sync starts right after; other systems are configured per instance. Most of the elapsed time is deciding which objects and filters to bring in.