One backend.
One database.
Many modules.
Nortra is built so a new module — Listr AI today, Missed Call AI next quarter, Review AI after that — can be turned on for any client without forking a codebase, standing up a database, or rebuilding billing. The infrastructure is the product.
Workflow Engine
The six workflows that run Listr AI today: lead intake, AI SMS reply, inbound matching, stop-on-reply, 24-hour follow-up, and agent notification. Each is configured per client, observable, and safe to update without downtime.
AI Reply Layer
OpenAI primary, Claude fallback. Every prompt is tuned per client. The moment a lead replies, the human takes over — there is no black-box generation pretending to be your agent.
Unified Inbox
SMS and email stitched into one conversation per lead. Threaded, append-only, and tagged with channel and source. Inbound webhooks are signed; replies trigger stop-on-reply within the same transaction.
Workspace Isolation
Each client gets their own workspace. Their data is theirs — no cross-workspace reads, no cross-workspace writes. This is what every production SaaS does. We enforce it in the database, not just the application layer.
Conversation Audit
Every inbound and outbound message stored. Append-only. Searchable by lead, by date, by source. When a client asks what we sent — there is an answer, not a guess.
Billing & Setup
Stripe is the source of truth for billing — Nortra mirrors state for the UI only. Setup fee plus monthly subscription per tier. Transparent invoices, no surprise overages.
The right pieces, wired together properly.
We use proven, boring infrastructure — and put our work where it actually matters: prompts that respond like a real human, follow-ups that fire on time, and a system that does not drop messages.