The agent fleet

AI agents that do the work — and earn your trust, one level at a time.

Six specialised agents. One autonomy dial. You decide how far each one goes — and every step is shown, scored, and stoppable.

Agent ConfigCloser
v2.4.1
Pause all agent outbound
Draft follow-up
Ceiling L2
L0Off
L1Suggest
L2Draft
L3Auto
L4Full

Closer drafts the reply and waits for your approval. Money & compliance actions can't exceed L2.

Dry-runNo actions takenSimulated 0s ago
  1. Scout would auto-ack new leadL3High

    Inbound from Priya Sharma re: 3BHK Whitefield — acknowledgement ready to send

  2. Scout would qualify via chatL2 draftMedium

    Budget ₹1.2 Cr · timeline Q3 · 2 dependents — qualification chat drafted

  3. Closer would draft follow-upL2 draftLow

    Personalised follow-up with Prestige Lake Ridge brochure — awaiting your approval

3 actions simulated · awaiting approval
Autonomy dial

The moat: how PropArch earns — and caps — agent autonomy.

Every agent action maps to a level. You set the ceiling. Levels three and four are locked for money and compliance — always.

The autonomy dial

Five levels. You hold the dial.

  1. L0
    Observe — watches activity, no outbound actions
  2. L1
    Suggest — drafts recommendations, you decide
  3. L2
    Draft & approve — agent drafts, you approve before any action executesCeiling for money & compliance
  4. L3
    Auto — executes autonomouslyLocked for money & compliance
  5. L4
    Full-auto — end-to-end autonomous executionLocked for money & compliance
The hard L2 ceiling

The rule that never bends.

An agent can draft a payment receipt — it can never finalise one. It can compute a payout — it can never release it.

Money and compliance actions are hard-capped at L2 across every agent, every action, every time. No override, no exception.

Dry-run

Preview before it's live.

See what an agent would do on a real lead before it's live. Every action in the simulation is identical to what would execute — no surprises, no side-effects.

No actions taken
Why-trail · confidence · kill-switch

Every action is explained, scored, and stoppable.

  • Why-trail: every action links back to the data that triggered it
  • Confidence score: High / Medium / Low on every recommendation
  • One-tap kill-switch: pauses all agent outbound instantly — no waiting
Agents

Meet your six-agent fleet

Three agents are live today. Three more are on the roadmap — each with a defined job, a clear autonomy ceiling, and hard stops at the RERA and GST gates.

  • SScoutLive

    Qualifies every lead in under 30 seconds

  • CCloserLive

    Drafts the next best action — you approve

  • LLedgerLive

    Assembles bookings, gated by compliance

  • CoConnectComing soon

    Keeps channel partners in sync

    Will keep your channel-partner network automatically up to date on inventory, pricing, and deal status.

  • PPulseComing soon

    Reads demand and prices to win

    Will surface demand signals and pricing insights to help you set competitive asks.

  • CnConciergeComing soon

    Guides the buyer after handover

    Will guide buyers through post-handover steps — snag lists, documentation, and maintenance scheduling.

Dry-run mode

See exactly what an agent would do on a real lead — before it's ever live.

Every agent action is previewed in full before it executes. No surprises, no regrets — just a transparent trace you review and approve.

Ask PropArch…
DRY-RUN · no actions takenSimulation mode
SScout

Priya Sharma · 3 BHK · ₹1.2 Cr · Baner, Pune

From 99acres · ready to move · home-loan pre-approved

Agent reasoning trace
  1. SScoutHigh

    Matched 3 live units in budget range

  2. SScoutHigh

    Buyer replied on WhatsApp in 6 min — high intent signal

  3. CCloserMedium

    Viewed floor plans twice · BANT score 94 — escalating to Closer

  4. CCloserHigh

    Saturday 11 AM slot free — optimal for site visit

Proposed actionSimulation only

Would draft a site-visit invite — Sat 11 AM

Simulation only — nothing was sent

Book a demo — we'll run the fleet on one of your real leads in dry-run, live.

Book a demo
Live today

Three agents, live now

Each agent has a defined job and a hard autonomy ceiling. Here's exactly what each one does — and where it stops.

SScoutLive

Acknowledges and qualifies every portal lead in under 30 seconds.

<30 s(design target, not a guaranteed SLA)

Scout reads the incoming lead, sends an instant acknowledgement, and scores it against your active project criteria — all before a human has had a chance to open the notification.

GuardrailScout escalates immediately if a lead shows buyer-intent signals that exceed its L1 qualification ceiling. A human broker takes over from that point.
CCloserLive

Drafts the next best action — you Approve, Edit, or Dismiss.

Based on conversation history and deal stage, Closer surfaces a ranked suggestion: follow-up message, site-visit invite, counter-offer framing, or escalation. Nothing leaves your CRM until a broker acts on it.

GuardrailCapped at L2 on anything outbound that touches money — pricing, token amounts, or payment references always require explicit broker sign-off.
LLedgerLive

Assembles a whole booking into one approval gate.

Ledger collects buyer KYC, unit selection, token amount, and payment schedule into a single structured package — then holds it at the gate until a broker confirms. Every field is validated before the gate opens.

RERA verifiedToken ≤10%70% escrowGSTTDS
GuardrailNo booking confirmation is sent to a buyer until all five compliance checks pass and a broker explicitly confirms through the gate above.
No black box

Signals → inference → 0–100 confidence

Every agent decision is a traceable chain. You see exactly what Scout read, what it inferred, and how confident it was before acting.

WhatsApp reply in 6 minHigh-intent signal
Floor plan opened ×2Engagement signal
Home-loan pre-approvedBuyer readiness signal
Inference

Buyer is visit-ready — BANT score 94, escalating to Closer

Confidence94 / 100
High

Below 0.7 confidence, Scout asks a human.

Category gap

This isn't a chatbot.

A Copilot suggests text inside a chat box. An agent fleet executes a workflow — within per-action guardrails and a server-side compliance gate — whether you're watching or not.

Copilot / chatbotSuggests
  • Sits inside a chat box — waits for a prompt
  • Drafts text; a human copies and pastes it
  • No awareness of live inventory, RERA, or hold state
  • Stops when you close the tab
PropArch agent fleetExecutes
  • Autonomous workflow — acts within guardrails you set
  • Per-action guardrails + server-side compliance gate
  • Inventory-aware, RERA-aware, full audit trail
  • Runs 24 × 7 across your entire pipeline

Platforms like Sell.Do, Anarock, Square Yards, LeadRat, DaeBuild, and Zoho CRM are built around lead records and dashboards — they do not have an executing agent layer with per-action guardrails and a server-side compliance gate. That gap is what PropArch is built to close.

See it live

Watch your fleet run a dry-run on your real leads.

Connect your lead source in minutes. The fleet dry-runs every agent step — Scout qualifies, Closer drafts the next action, Ledger checks compliance — without touching a single live record. You decide when to go live.

Book a demo