Working demo · AI sales operations

Propadya AI Sales Command Center

A publicly inspectable working demo connects intake, qualification, explainable opportunity matching, follow-up drafting, buyer briefs, and manager visibility in one fictional-data workflow.

Status: Working demo
Date: 18 June 2026
Role: Arian Asadi — product architecture, full-stack implementation, AI workflow design, data modeling, localization, testing, and release hardening
Context: Self-initiated working demo branded Propadya. Its public deployment contains fictional people, opportunities, pipeline values, SLAs, and outcomes; it is not presented as a live client deployment or a source of measured business results.
Stack / direction: Next.js 15 · React 19 · TypeScript · Prisma · PostgreSQL · Vitest · OpenAI Responses API (optional)

Arian Asadi / self-initiated demo

Problem and constraints

High-value property inquiries lose momentum when qualification, opportunity matching, follow-up, buyer briefs, and manager quality control are split across unstructured conversations and disconnected tools.

Every visible person, property, value, SLA, and pipeline result must remain clearly fictional
AI may recommend, rank, summarize, or draft, but it may not send, book, change inventory, or make legal, citizenship, yield, or client commitments
The useful demo path must still work when no OpenAI credential is configured
Demo roles and process-local rate limits cannot be presented as production authentication or abuse control

Key decisions

Make AI recommend the next move while salespeople keep every commitment.

01
Map the lead lifecycle into one executive operating surface instead of another disconnected assistant
02
Model consent, ownership, opportunities, matches, conversations, follow-ups, AI logs, and activity history as durable records
03
Use structured AI outputs with an explainable deterministic fallback and visible human-approval state
04
Keep messaging, CRM, calendar, and transcription providers behind explicit non-live adapter boundaries
05
Ship English and Turkish interfaces, release documentation, focused tests, and a guided five-minute demo path

Architecture

From inquiry capture to an approved, auditable next action.

1Validated lead intake
2Durable sales record
3Explainable opportunity ranking
4Approval-gated AI drafts
5Manager control and audit trail

Evidence currently available

Live guided demo

The public executive-demo and dashboard routes render the fictional-data workflow, visible approval boundary, lead priorities, pipeline, follow-up control, and manager view without browser errors.

Source: Propadya public demo · 2026-07-18

Reproducible release checks

At commit b6de644, lint, strict typecheck, nine focused tests, and the Next.js production build passed; the production dependency audit reported zero known vulnerabilities.

Source: Private repository verification at b6de644 · 2026-07-18

Durable operating model

The reviewed Prisma schema and initial migration model users, sales representatives, leads, opportunities, matches, conversations, follow-ups, AI logs, activity, consent, timestamps, uniqueness, and operational indexes.

Source: Private Prisma schema and migration · 2026-07-18

Controlled AI boundary

Structured qualification, matching, summaries, briefs, and follow-up drafts use safe fallbacks; external adapters remain non-live and follow-up output requires human approval.

Source: Reviewed AI modules, schemas, prompts, and adapter contracts · 2026-07-18

Confidentiality: The live workspace contains fictional demo records only. Names, properties, values, response times, pipeline totals, meetings, and won amounts are illustrative product data—not customer information or achieved business metrics.

Contribution and transparency

Arian's contribution

Product framing and executive workflow
Information architecture and multilingual interface
AI fallback, approval, and adapter boundaries
PostgreSQL data model, tests, and release hardening

Assistance and platforms

AI-assisted development tools used under Arian Asadi's direction and review
Next.js, Prisma, PostgreSQL, and the optional OpenAI provider path

No client deployment, external collaborator, or live customer-data contribution is claimed for this self-initiated demonstration.

Next verified step

Replace demo roles with real authentication and record-level authorization, then add shared rate limits, provider threat models, consent and retention policy, monitoring, backup and restore rehearsal, and a permissioned staging integration before any real client data.

Back to public proof Open working demo Discuss a related system