SafeFlowSME AI Challenge · Neuron Dynamics

Secure the Data,
Streamline the Flow.

SafeFlow lets SME staff ask about sales, finance and HR data in WhatsApp, Microsoft Teams, Google Chat or Slack, and get a trusted answer, a chart and the metric definition right in the chat. Permissions are checked before every answer.

TrackOperations & ProductivitySME AI Challenge
CompanyNeuron Dynamics LimitedHong Kong
Available in
  • WhatsApp
  • Teams
  • Google Chat
  • Slack
Start with the 2-minute video ↓
01Overview video01 / 10

Two minutes: the AI at work, and the safeguards that keep it in check.

Watch on YouTube ↗

02The problem02 / 10

SMEs need answers from their data. But reports take days, and sharing sensitive data with AI tools is risky.

Scattered data

Sales, accounting and HR sit in different spreadsheets and systems.

Slow reports

A simple question can wait days for someone to compile a report.

Hard to self-serve

Store managers and other non-technical staff cannot query data themselves.

Sensitive data at risk

Pasting sales, finance or HR data into public AI tools can expose it, so many SMEs avoid AI altogether.

03Our solution03 / 10

Ask in chat. Get a trusted answer.

In the apps they use

WhatsApp, Microsoft Teams, Google Chat and Slack. Nothing new to install or learn.

Knows your metrics

Answers use the metric definitions your data owner approved, so everyone counts the same way.

Answer, chart, definition

In English, Cantonese or Chinese, following the language of the question.

Safe by design

Permission-aware, read-only and fully audited.

04How it works04 / 10

Five steps from question to answer, with a safeguard at each one.

Ask

A question in plain language, in any supported chat app.

User & organisation verified

Understand

The AI maps the question to an approved metric and time range.

Manipulative prompts blocked

Retrieve

Reads only data this user may see, from one secure workspace.

Role-based access

Analyze

Runs a read-only query. Sensitive values are masked before any external AI call.

Read-only, sensitive values masked

Answer

Returns the answer, a chart and the definition, and logs the request.

Audit trail
05Product05 / 10

The same question in WhatsApp and Teams.

“What were monthly sales and growth over the last 12 months?”

SafeFlow in WhatsApp on a phone: the sales chart arrives in the chat, followed by the analysis
WhatsApp · chart and analysis in the chat
SafeFlow in Microsoft Teams: the answer card with key takeaways and monthly detail
Microsoft Teams · the same answer as a card

Screenshots from the live product. The data is a fictional sample workspace.

06SME use case06 / 10

A Hong Kong F&B chain with 5–10 stores.

Operations & Productivity Food & beverage 5–10 stores · owner, finance, HR, store managers
BeforeWith SafeFlow
The business owner has to ask staff for reports.They ask on WhatsApp and get the answer in the chat.
Finance compiles ad-hoc reports by hand.Routine questions are answered without a report request.
Finance and HR data is shared by spreadsheet.Each person sees only what their role allows.
Users

Owner, head-office finance, HR and operations, and store managers on mobile.

Scope

Sales, operations, accounting and HR (authorised users only). Read-only, no write-back.

Approach

Connect existing Excel, Google Sheets, OneDrive or database sources, agree metric definitions, publish a versioned workspace.

Operating model

We set up and maintain the workspaces. The client’s data owner approves metrics and each new version.

Deployment

First answers in about 2–4 weeks; full rollout in about 4–8 weeks.

How it is used

Store managers ask on WhatsApp; head office asks on WhatsApp or Teams.

What changes

Time to answerSeconds instead of hours or days
Reporting effortLess manual report preparation
Ad-hoc requestsFewer report requests to head office
Self-serviceStore managers answer their own questions
07Security & governance07 / 10

Safe by design, and approved by the client.

SafeFlow in WhatsApp declining an HR team member's supplier question because only purchasing can see that data
Role-based access in action. A member of the HR team asks about supplier data that only purchasing can see. SafeFlow declines and explains why.

Access & sign-in

Only approved Microsoft 365 organisations and verified WhatsApp numbers can ask. Data owners connect sources with their own Google or Microsoft account.

Role-based access

Permissions are checked before every answer. HR and finance data can be included without opening it to everyone.

Data-source control

Only sources the data owner selects are connected. Each conversation reads one isolated workspace.

Input guardrails

Manipulative prompts and pasted sensitive data are blocked before they reach the AI.

Data protection

Credentials stay in an encrypted vault in the client’s environment. Sensitive values are masked by a private model, and client data is never used to train models.

Audit & approval

Every question is logged. The owner signs off data scope and roles, the data owner approves each version, and failed updates roll back.

Designed in line with good practice under Hong Kong’s Personal Data (Privacy) Ordinance. Larger or regulated clients can add an IT security review and deploy in their own cloud.

08Implementation08 / 10

Up and running in weeks.

Weeks 1–4

Onboard

  • Connect your existing data sources
  • Set up roles and approve metric definitions
Weeks 4–8

Roll out

  • Staff ask in their own chat app
  • Add departments and data sources
Ongoing

Grow

  • New workspaces as the business grows
  • Enterprise features and support

Pricing model: a one-time setup fee plus a monthly subscription, with an SME tier and an Enterprise tier (more users and sources, advanced permission mapping, security review support). Pricing scales with users, data sources and workspaces.

09Team09 / 10

Enterprise data-protection engineers, plus Hong Kong go-to-market.

Cheng Yang

Cheng Yang

Project lead · Product & engineering

6+ years as a Senior Engineer at Dell EMC on an enterprise data protection SaaS platform. Previously PwC and UBS. MS Computer Science, Georgia Tech.

Harris Chan

Harris Chan

Go-to-market

5+ years in Hong Kong SaaS and FinTech business development, 1,000+ merchants managed. Global Payments, OpenRice, FiberAPI.

Yuanyi Liu

Yuanyi Liu

Platform architecture

15+ years; Senior Principal Engineer at Dell Technologies. Architected enterprise data pipelines for three major products. Tongji University.

Yuefeng Li

Yuefeng Li

Backend & infrastructure

6+ years as a Senior Engineer at Dell EMC. Backend, Kubernetes deployment and concurrency. BS & MS, Shanghai Jiao Tong University.

In their previous roles, three of our engineers earned 30+ patents granted in China and the US.

10Contact10 / 10

See SafeFlow with one of your own questions.

Book a 30-minute demo on Google Meet, or reach us directly.

Book a demo Visit safe-flow.tech