StaffOS

The AI sales rep that
listens before it talks.

Lisa speaks to every new lead within 30 seconds of form submit. Qualifies them in their language. Then she stops, watches, and only re-engages when a real-world signal — a page revisit, an ad re-click, a question answered by your new KB article — gives her something to say. No timer-based drips. No "just checking in." No template blasts.

Hire Lisa See Lisa in action
L
Lisa
Inbound Sales Agent
Live
Hi, saw your ad. How much is the recovery package?
Matching the ad this lead clicked
Pulling current package pricing
Hi Mei! You came through our Recovery Package ad. The six-session package is RM 1,200, and first visits include a posture assessment. Is this for yourself or someone else?
For my dad. He has knee pain, been going on for months.
Checking this week at the KL branch
For knee pain we start with the assessment so the therapist can tailor the sessions. I have Thursday 3pm or Saturday 11am in KL. Which suits him better?
Saturday. Can you confirm?
Booking Saturday 11am
Done. Saturday 11am at KL, under your father's name. I have sent the address and what to bring. Aisha from our team will look after him from here, and she has everything you told me.
Passing the lead to a rep

Handed off to a human

Booked, and passed to Aisha with the transcript, the ad it came from, and what Mei said about her father.

Ready
Real-time

Conventional AI nurture inherited drips from a pre-LLM era.

When marketing automation tools were built, bots could not remember anything, so the industry substituted a schedule for memory. The result is a stack that sends "Hi {name}, just following up!" on a timer regardless of what happened on the lead's side. Lisa rebuilds the bottom of the funnel. She listens for real-world signals — page revisits, ad reclicks, KB updates that resolve a prior objection, time-of-day patterns the lead taught you about themselves — and only re-engages when one of those signals justifies the cost of speaking up.

Listen. Detect. Engage.

Listen

After the first outbound, Lisa enters listening mode. Zero scheduled follow-ups. A signal bus indexes events per lead — page revisits, ad re-clicks, read receipts, sentiment shifts, KB resolution flips.

Detect

Each signal is interpreted. Not every signal triggers outreach. A page revisit during the lead's active hours is engagement; the same revisit at 3am when the lead never replies at 3am is a wait. A "stop pls" reply is an instant opt-out.

Engage

When a signal justifies outreach, Lisa composes using the full context she has — the prior conversation, the campaign source, the KB entry that resolved the objection, your brand voice. Every message has a reason. You see the reason next to the message.

What Lisa listens for, and what she does about it.

Lisa's nurture engine subscribes to a signal bus. These are the six signal types that drive most of her re-engagement decisions today.

Page revisit

When it fires

Lead opens your landing page again, days after going dark

What Lisa does

"Saw you back on our site. Happy to walk you through the floor plans live."

Ad re-click

When it fires

Lead clicks your Meta ad a second time

What Lisa does

Match the new ad creative: "You saw our Spring Launch ad again. Want the unit specs?"

Lead reply

When it fires

Lead replies to a previous message

What Lisa does

Pick up the thread with full context. Lisa remembers objections raised three weeks ago.

Objection resolved

When it fires

You publish a KB update that addresses an objection a lead previously raised

What Lisa does

"Last time you mentioned X was a blocker. That has just changed..."

Read, no reply

When it fires

Lead read your last message but did not reply

What Lisa does

Record only. Adjust the engagement model. Do not push.

Negative sentiment

When it fires

Lead replies with dismissive tone ("stop pls" / "leave me alone")

What Lisa does

Exit sequence. Flag for human review. PDPA opt-out logged the same minute.

From ad click to revenue, closed.

StaffOS does not stop at the conversation. It follows the whole journey and reports what actually drove revenue.

  1. 01

    Ad click

    A lead taps your ad and lands straight in chat.

  2. 02

    Conversation

    An agent qualifies and answers in seconds.

  3. 03

    Booking

    It books the meeting on a real calendar.

  4. 04

    Conversion

    The outcome goes back to your ad platform.

  5. 05

    ROI

    You see spend against revenue on one screen.

Most "AI nurture" is drip with better copy. Lisa is something else.

FeatureDrip cadence toolsLLM-rephrased dripEnterprise workflowLisa
Cadence modelTimer-basedTimer-based (smarter copy)Workflow nodesSignal-based
Off-script reply handlingBreaksFalls back to cannedRoutes to humanReasons + responds
WhatsApp template strategyBlast at +25hBlast at +25hBlast at +25hFire only on signal
Per-campaign autonomyNoneAll-or-nothingNoneEarned per campaign
Meta CAPI loopForm-fill onlyForm-fill onlySometimes manualLead / Schedule / Purchase / LDU
Bahasa Melayu / IndonesiaSomeEnglish-firstAdd-onNative
Pricing modelRM-friendlyUSD 3 to 15k/monthEnterprise contractRM-friendly

Lisa earns autonomy per campaign, not all at once.

Lisa does not graduate to autonomous send on day one. Each campaign accumulates its own confidence track — accept-verbatim rate, edit-distance, discard rate. When a campaign's accept-rate crosses an operator-set threshold over a minimum sample, Lisa graduates to autonomous send on that campaign. Compliance-heavy verticals stay in approval mode forever, by design.

01

Draft

Every outbound message routes to operator approval. Edits become training data for the campaign.

02

Supervised

High-confidence drafts auto-send on the campaign's graduated channels. Operator reviews exceptions only.

03

Autonomous

Lisa sends without review on graduated campaigns. Operator audits a sample.

Questions sales managers ask first.

What if I want a drip cadence?
You can opt in to it per campaign. A silence-break signal can fire after N days of silence as a minimum-floor catch net. It is off by default. Verticals with long decision cycles (insurance, property) sometimes benefit from a baseline cadence — those tenants flip it on per playbook.
How does Lisa know the lead's reply window?
A per-lead reply-window model. Lisa tracks every reply timestamp; the active window emerges from the lead's own behavior. New leads default to the campaign cohort's median reply hours.
What if Lisa gets confused mid-conversation?
Two safety layers. First, every customer-visible message routes through the campaign's autonomy tier — new campaigns start in approval mode. Second, every tool Lisa uses is preconditioned. She cannot send a booking link without captured Interest and Timing signals. The tool itself refuses.
What about cold outbound prospecting?
Different product. Lisa is for inbound — tenants who already have a paid lead source (Meta ads, web forms, walk-in WhatsApp). Cold prospecting is a separate future product. The deliverability and signal model are fundamentally different.
Bahasa Melayu? Singlish? Bahasa Indonesia?
Native. Lisa mirrors the customer's language and register. The same agent handles a formal-English property lead from Penang and a Manglish casual reply from KL on the same campaign and shifts register accordingly. Speed-to-lead is in their language, not yours.
How much does this actually save on WhatsApp templates?
For a tenant doing 200 leads per month with a 4-touch drip cadence outside the 24-hour window, the conventional model fires roughly 600 template sends per month. At Malaysia rates as of July 2025, that is RM 300 to 480 per month. Lisa typically fires 40 to 80 template sends per month — RM 20 to 65. Your actual number depends on lead volume and signal density.
Can I see what signals fired and why?
Every signal is logged with kind, payload, timestamp, the engine's decision, and the message that was sent (or not). The operator dashboard has a signal log per lead. Audit-ready.
What if my brand voice is unique?
Lisa inherits your brand persona — voice notes, example replies, signature line. You can edit it conversationally in the operator twin chat: "Drop the exclamation marks and use shorter sentences when replying to property leads." She learns and updates the prompt.

Speed-to-lead in 30 seconds. Then she listens.

Lisa onboards in a day. Your Meta ads, your CRM, your brand voice — connected, grounded, and converting by tomorrow.

Hire Lisa Meet Sam, our Customer Support