In Semi-Auto, only the specification file is parsed. Images/videos are uploaded for later steps, but are not OCR’d here.
No files loaded yet.
Raw spec preview
Detected listing details
Auto-filled from spec files and/or OCR. Anything still empty is flagged so you can complete it.
The spec does not say where this property is — tell me once and I will remember it for this property
Nearby amenities & description:
Measured from OpenStreetMap — check before using
Draft description — review before using
AI Content Studio
Turns the listing facts from step 1 into ready-to-publish copy — pick languages and tones and generate. You can frame the AI output with your own custom opening and closing lines.
visibility
Visibility engine writes: an answer-first opening line (quotable by AI answer engines), key facts, an FAQ pair set, geo/entity terms, keyword phrases for portals, and platform-tuned hashtag counts.
✓ Saved
Hashtags — appended to the post at publish time
Tip: tick several languages to get the whole write-up in each language, or several tones and the engine will weave them into one natural piece. Your header/footer and ticks are remembered per user.
Auto-enhance for advertising
Enhancement Studio (v1): scene-aware exposure & white balance, denoise, straighten, style presets — plus sky enhancement and disclosure-stamped AI-grade edits that wait for your approval. Tune processing defaults in Settings → Enhancement Studio.
Default mode enhances each photo individually.
These four lines are what make the poster read as an advert. Price is drawn largest, the headline beneath it, then location.
Sky enhancement and twilight conversion alter the appearance of the photo, so each one waits here. Approve to make it the active enhanced image, or reject to keep the previous version.
Branding & overlays active
Every enhanced photo can carry up to four branding elements — a text watermark, your logo/sticker, a price badge and a selling icon. Each picks a zone on the 3×3 grid, carries its own size, and can be switched off without losing its settings. The preview is composited by the same engine that burns the final image, so it is literal, not an approximation.
Logos you have used before — tap to reuse (it drops into the next free corner). Uploads above are saved here automatically.
Pick a selling icon — it drops into the next free corner:
Apply branding to
Enhancement still runs on every image — only the branding stack follows this choice.
Layout preview
Building preview…
dashed amber = the area that element occupies
Assemble the promo video
Pick photos, choose a voice, tune motion & format — Enhancement Studio v1 renders Ken Burns motion, styled captions, optional music ducking and a narration walk-through with a viewing invitation. Long renders run as a job with progress.
1 · Choose photos (click to select — the order you click IS the video order)
Auto-selectNo photos loaded
Sequence — reorder to architect the flow; this is the exact playback order
2 · Narration script editable before render
Auto-generated from listing facts — edit any line; blank lines become silent slides. One line per selected photo.
3 · Voiceover, motion & format
Presenter voice reference (optional)
Upload a ~30s recording of your preferred presenter — the app analyses its pacing and auto-suggests the closest natural voice. Enable the ElevenLabs engine (⚙ settings) to clone the exact timbre as a custom narrator.
Voice engine
Voice sample — upload & analyse
Upload ~30s of the presenter's voice. The app derives pacing/tone and suggests the closest built-in voice; with the ElevenLabs engine enabled it also creates the clone.
The script above is generated from the listing facts and the AI copy in the Content Studio — edit it freely before rendering. Progress updates while the job runs.
4 · Your video
Preview & publish
Everything a portal listing needs, tabulated in one place, with the enhanced photos and video ready to attach — then push to connected channels.
Listing summary
Listing description
Load a listing to see its enhanced description here.
Post preview as published
Exactly what will be posted: your custom header, the AI copy, then your custom footer. Edit them in Content Studio — this block updates automatically.
Per-channel captions
Each channel gets its OWN caption, rebuilt to that channel's length from your researched limit table — the hook and strongest selling point inside what shows before "See More", and never over the platform's cap. PropertyGuru / iProperty get no contact details in the copy. Nothing is published by this step.
Press Prepare to build one caption per channel for review.
Enhanced photos
Video
No video generated yet — go to step 4.
Channels
Leads & Appointments
Every enquiry, scored automatically by urgency, source, and budget fit — with a follow-up action suggested on the spot.
Simulate an inbound enquiry
Pipeline
Appointments
Project Launch
A launch is a SEQUENCE — Pre-Launch → Launch Week → Post-Launch — never a single campaign. Track the phase checklist, absorption and cost-per-outcome here. Pricing, budget and incentive changes are recommendations only, never auto-applied.
Launch
Create a launch
✓ Created
Unit inventory (feeds absorption)
Spend & attribution cost-per-outcome
Agility Settings
This actually governs the background scheduler — set it to Conservative and automated follow-ups pause.
Guardrails
✓ Agility profile applied.
What changes at this level
Conservative pauses the background follow-up sweep entirely. Balanced and Autonomous both run it every 90 seconds, but Autonomous never waits on quiet-hours for Hot leads (Balanced still does if the quiet-hours guardrail is on).
Conversations
Every WhatsApp conversation by number — the full thread with media, plus what the lead actually wants: buyer segment, market tier, budget band and interest factors.
Loading…
Lead profile
Thread
Pick a conversation on the left to read the thread and the lead's segment and interests.
Replies
Semi-auto: the agent composes a reply to every enquiry, and nothing is sent until you press Send. Set a channel to semi to start queueing drafts here; draft keeps today's behaviour (the draft only appears in the conversation thread).
Reply mode per channel
Loading…
Auto is shown but not enabled yet — the backend treats it as semi so nothing can be sent unreviewed.
Waiting for your approval
Loading…
Settings
Super Admin surface. Pick a user to open their Profile & Access and change anything on their account — role, password, recovery contacts, agent profile and saved presets. Their own login keeps working exactly as before.
Users
Create user account
Add an ordinary user, hand them the temporary password, then they change it in their own Profile & Access.
Created
Configure where published listings go. Add credentials per channel, enable/disable, or remove. Test each connection before using.
Channel inventory
Configure channel:
Add a new channel
Add, remove, or switch the AI providers used for listing copy generation. One active provider at a time. Each provider holds its own API key + model. Remove means it no longer appears in the picker.
Providers
Add AI provider
✓ AI settings saved.
Professional photo & video enhancement defaults. Corrective edits (exposure, denoise, upscale) apply automatically; AI-grade edits (sky enhancement, twilight) stamp a disclosure label and wait for your approval before they become the active image.
Photo enhancement
Feature toggles
✓ Saved
Video enhancement
Running image (silent slideshow) — all photos in one MP4, no narration
✓ Saved
Daily usage & caps
Voice cloning & presenter settings. Capture a presenter's voice, record the consent that authorises its use, and choose the engine, languages, residency and access rules. Consent is always required and cannot be switched off.
Engine
✓ Saved
Presenter profiles
A profile is a captured voice + its rights record. It cannot be used until consent is recorded, and revocation takes effect immediately.
1 · Capture a sample
✓ Created
Consent & rights record
Required before any synthesis. Keep the signed agreement reference — this is the audit trail.
✓ Saved
Visible speaking presenter
When off, narration plays over the stills exactly as today. When on, the audio side is unchanged and the visual provider is applied once wired.
Languages
Which languages the cloned voice must speak. The provider must support cloning and synthesis in each one.
Data residency
Where audio processing is allowed. This is enforced: on-prem or third-party-disabled refuses managed APIs.
Who may generate Super Admin
Restrict voice generation to a chosen list of users. The Super Admin is always allowed.
Output & volume plan
✓ Saved
Agent profile (auto-fills name/contact across listings, videos and publishing), your login credentials, recovery options, and the in-app brand.
Agent profile
✓ Saved
Login & recovery
Initial sign-in: username vincentkoay / password 123456 — change it here after first login.
user
✓ Updated
My saved presets
Per-account preferences: your narrator voice and video defaults are remembered for this account on every device and restored automatically.
✓ Saved
In-app brand
✓ Applied
Inbound message log
● live
Read-only. Each inbound WhatsApp message received via the Meta webhook is captured here for quick monitoring. The real record lives in Lead → Pipeline.
0 messages
Missing details
Only what could not be read from the spec. Type a value, tap an option, or tap Speak once to start and again to stop — whatever field is selected receives it.