groupers.ai
groupers.ai · 投資簡報groupers.ai · Investor Pitch

同你一齊營運間公司嘅 AIThe AI that runs the company with you

公司內外嘅溝通同文件,佢幫你打理 —— 每一句出街,都係你批先出It manages your communication and documents, inside and outside the company — and every message leaves only on your yes.

簡報連現場示範 · 20 分鐘內Pitch + live demo · under 20 minutes

市場痛點The Problem

香港中小企:成盤生意喺 WhatsApp 度,但得你一個人覆Hong Kong SMBs run on WhatsApp — and one person answers everything

好多老闆每日喺呢啲溝通同文書度用多過 3 個鐘 —— 呢啲係佢最貴嘅時間Many owners spend over 3 hours a day on this communication and paperwork — their most expensive hours
Easy-to-use AI
解決方案The Solution

三步,搞掂Three steps, done

1

任何人搵你Anyone messages you

客人、師傅定供應商 —— 訊息照舊入你嘅 WhatsAppCustomer, technician or supplier — it lands in your WhatsApp as usual

2

AI 俾三個答案你揀AI drafts three replies

用你教過嘅價錢、規矩、口吻寫好Written with your prices, your rules, your tone

3

你覆個「1」You reply "1"

就咁送出 —— 你唔撳,佢一個字都唔會出街It sends — if you don't approve, not a word leaves

信任設計Trust by Design

四個保證,唔係口講Four guarantees — built in, not promised

🔒

唔會自己出街Never sends alone

你未撳,佢就唔會送 —— 一句都唔會Until you approve, nothing is sent — not one line

唔識就問你Asks when unsure

唔會靠估亂答客人,寧願問返你It never guesses with your customers — it asks you instead

🚫

唔會亂搵人Never cold-calls

唔會 cold DM、唔會群發 —— 新聯絡人一定經群組或邀請連結入No cold DMs, no broadcasts — new contacts only join via a group or invite link

📖

記住嘅嘢你睇得晒Its memory is open

每一句都係你教嘅 —— 冇黑箱Every line was taught by you — no black box

呢啲唔係設定,係佢天生做唔到壞事嘅設計 —— 現場可以試。These aren't settings — the system is built so it can't misbehave. Test it live today.

Working Agent
產品示範 ①・流程動畫Product Demo I · Walkthrough

就算要問內部同事,一樣搞掂Even when it needs a teammate's answer, it runs the relay

Working Agent + designed SOP
行業應用Industry Applications

你嘅行業,佢都幫到手It works in your industry

🧹 家務助理Housekeeping 睇動畫Play

屋主要求多籮籮 —— 教一次,AI 次次同姐姐逐項對Detailed household rules — teach once, the AI checks every point with the helper, every time

📦 貿易・採購Trading 睇動畫Play

一單貨要問幾間供應商 —— 價同期日日變Every order means asking several suppliers — prices and lead times change daily

💻 IT・工程IT & engineering 睇動畫Play

客戶亂報 bug —— 工程師要嘅係清楚工單Clients report bugs in a mess — engineers need clean tickets

🍜 餐廳・服務業Restaurants 睇動畫Play

夾更、訂位改期 —— 全部喺 WhatsApp 度追Rosters and booking changes — all chased over WhatsApp

邊個行業都係同一條規矩 —— 出街嗰一下,永遠係你批Every industry, one rule — the send is always your call.

Human-in-the-loop
核心能力Core Capabilities

唔止識覆 —— 識問、識記、識做嘢Not just replies — it asks, remembers, and does

唔識就問你Asks when unsure

「needling 後可唔可以游水?」—— 冇資料寧願問返你,唔會靠估亂答客;你答完佢即刻記低"Can I swim after needling?" — no data on file, so it asks you instead of guessing with your customer; your answer is saved on the spot

📌

教一次記一世Teach once, kept forever

WhatsApp 打句「記住,兩房減到 780」—— 之後所有擬稿自動用新價Type "remember: the 2-bed is down to 7.8M" — every draft after that uses the new price

📥

群組摘要Group digests

「總結 業主群」—— 51 條未讀變三個重點,邊條要覆都話你知;佢喺群入面只聽唔講"Summarise the owners' group" — 51 unread become three bullets, flagging what needs your reply; it never speaks in the group

📨

主動代辦Runs your errands

「同陳太講改咗十點」—— 佢寫好晒你批先發,把聲係你"Tell Mrs Chan we moved to 10am" — drafted for your approval, in your voice

行政嗰啲追嘢都係佢做:會計月結追單據、核數師要確認、資助死線 —— 追嘢跟嘢我哋做,盤數唔係我哋做。出街嗰一下,永遠係你批。It also runs the admin chase: month-end receipts, auditor confirmations, funding deadlines — we do the chasing, not the bookkeeping. And the send is always yours.

Applicable AI — saves your time
產品示範 ②・現場實測Product Demo II · Live

現場真機,同佢真人咁傾Live on the real thing

現場同我哋個平台 bot 真實對答:問佢產品嘢、睇佢行一次開通流程 —— 佢每個回覆要諗大約半分鐘,啱啱好夠我哋講解佢背後做緊乜。A live conversation with our real platform bot — ask it anything, watch it walk through onboarding. Each reply takes about half a minute of thinking — just enough time to explain what it is doing underneath.

Agent-to-Agent communication
願景Vision

當每間公司都有 groupers.ai:AI 同 AI 傾,人齋批When every business runs groupers.ai: AIs talk, humans approve

😎🤖 你(貿易行) You (trading co) 👔🤖 供應商 Supplier 🍜🤖 餐廳 Restaurant 🚚🤖 物流 Logistics 報價・夾期・落單 Quotes · schedules · orders 訂貨・對數 Ordering · reconciliation
  • 今日:AI 幫一間公司對人Today: your AI works your conversations
  • 聽日:你嘅 AI 直接同供應商嘅 AI 對盤 —— agent 對 agent 傾報價、夾期、落單Tomorrow: your AI deals with your supplier's AI — quotes, schedules, orders, agent to agent
  • 每一步出街,仍然係人批 —— 呢個係信任層,亦係我哋嘅護城河Every send still needs a human yes — that trust layer is the moat
  • 你而家教熟嘅 AI,就係你將來喺 agent 網絡入面嘅身分The AI you train today becomes your identity on tomorrow's agent network

用戶愈多,網絡愈值錢 —— 我哋做緊嘅係人・agent 商業互動嘅引擎,唔綁死喺任何一個通訊平台;WhatsApp 只係而家最就手嘅入口。Every user makes the network worth more — we're building the engine for human-and-agent business interaction, tied to no single platform; WhatsApp is just today's most convenient door.

市場時機Why Now

AI 到咗,但一般人用唔到 —— 呢個就係空隙The AI is here — most people can't use it. That's the gap.

🧠

AI 能力已經到位,但對大部分人嚟講唔易用 —— prompt、設定、新工具,唔係小老闆會學嘅嘢The capability has arrived, but for most people it's not usable — prompts, setup and new tools aren't things a shop owner learns

🎢

一般 AI 工具自由度太大 —— 要用得安心,反而要約束 AI 嘅能力:邊個可以講、幾時先出得街Most AI tools have too much freedom — to be trusted, AI needs constraints: who it may speak to, and when a message may leave

🔒

我哋就係做呢層約束 —— 所以最啱通訊密集嘅中小企:日日喺 WhatsApp 做生意、冇時間學新嘢、錯一句都蝕唔起That constraint layer is our product — built for communication-heavy SMBs: on WhatsApp all day, no time for new tools, no room for a wrong promise

潮水退咗,淨低嘅係有真客、真收入、真工作流嘅公司。When the hype tide goes out, what's left are companies with real customers, real revenue, real workflows.

市場機會Market Opportunity

香港做證明,世界先係市場Hong Kong proves it — the world is the market

36 萬香港中小企(官方數)—— TAM 錨HK SMEs (official) — the TAM anchor
~10 萬對客、通訊密集嘅細團隊(人口反核吻合)Customer-facing, communication-heavy small teams (population cross-check)
~1,000香港階段目標 ≈ HK$1,000 萬 ARR —— 驗證 + reference 完成HK-stage target ≈ HK$10M ARR — validation + reference complete
100–300灘頭:兩個行業,12–18 個月,逐間簽Beachhead: two industries, 12–18 months, signed one by one
🌍 國際先係真正嘅市場 —— 引擎唔綁平台,香港 playbook 直接複製出去International is the real market — the engine is platform-agnostic; the HK playbook copies straight out

背景:香港有 36 萬間中小企、~10 萬隊通訊密集細團隊 —— 1,000 間連 1% 都唔使,個目標係細到可以信Context: Hong Kong has 360k SMEs and ~100k communication-heavy small teams — 1,000 customers is well under 1%. The target is small enough to believe.

而且淨係香港:HK$1,000 萬 ARR 以一般 SaaS 倍數計,已經撐得起今輪估值嘅 10 倍以上 —— 出海係喺呢個底上面再乘。And Hong Kong alone: HK$10M ARR at standard SaaS multiples already supports 10x+ this round's valuation — international multiplies from there.

競爭格局Competitive Landscape

個個都話有 AI,點解係我哋Everyone claims AI — why us

替代方案Alternative問題The problemgroupers.ai
ChatGPT・一般 AI botChatGPT / generic AI bots個人工具 —— 幫你諗嘢寫嘢,唔係為公司溝通而設Personal tools — great for thinking and writing, not built for business communication為做生意嘅溝通而生 —— 人批先出Built for business communication — approval first
Slack・新協作工具Slack / new collaboration tools功能強,但小老闆唔會學新 toolPowerful — but small owners won't adopt new tools零學習成本 —— WhatsApp 本身Zero adoption — it is WhatsApp
SleekFlow・Omnichat 呢類 WhatsApp 商務平台WhatsApp commerce platforms (SleekFlow, Omnichat)做群發同客服 inbox —— 幫你「發得多」Broadcasts and support inboxes — built to send more我哋唔做群發 —— 做「你把聲、你批先出」We never broadcast — your voice, your approval

我哋嘅客係唔會自己搵新 tech tool 嗰批人 —— WhatsApp 係最快嘅入口。Our customers are exactly the people who never go shopping for new tech — WhatsApp is the fastest door in.

落地策略Go-to-Market

頭一批客,唔使周圍搵Our first customers are already picked

商業模式Business Model

一個核心引擎,兩種賣法One core engine, two ways to sell

標準版買量、訂製買深 —— 行業專家同時就係我哋嘅白牌批發網絡SaaS for volume, customisation for depth — the industry professionals double as our wholesale network.

天使輪融資The Ask · Angel Round

US$200K,換 10%US$200K for 10%

US$200K

天使投資額Angel investment

10%

股權Equity

US$2M

隱含估值(post-money)Implied valuation (post-money)

資金用途USE OF FUNDS
30%
30%
35%
30% 技術團隊人工30% tech team salaries 30% AI tokens・軟件・基建30% AI tokens · software · infrastructure 35% BD・市場推廣35% BD & marketing 5% 行政5% admin

30% 落 tokens 唔係浪費 —— 以今日 AI 工具嘅效率,呢筆錢買返嚟嘅係一年前 15 個工程師嘅生產力30% on tokens isn't overhead — at today's AI-tool efficiency, that spend buys what fifteen engineers' output cost a year ago.

多謝Thank You

groupers.ai —— 你批,佢做groupers.ai — you approve, it does

我哋唔係砌更叻嘅模型 —— 係砌「邊個可以講、幾時先出得街」嗰層。We’re not building a smarter model — we’re building the layer that decides who it may talk to, and when a message may leave.