Hexafusion Labs · AI research and development
Hexafusion Labs
Hexafusion Labs is where we build and test the artificial intelligence tools we eventually bring to our managed services and to clients across the industries we support. Everything described on this page is already running inside Hexafusion today, not a concept or a roadmap item.
What is already in production
These are the four AI projects running inside Hexafusion right now.
AI phone assistant on our own line
An AI assistant answers calls to Hexafusion. It recognizes existing clients and routes them appropriately, logs tickets, takes messages, and tells every caller plainly that they are speaking with an AI. Details on the AI phone assistant page.
Email threat triage
An AI model assists in classifying incoming email as spam, phishing, or legitimate. A person on our team reviews the classification before any action is taken on a client mailbox. The AI narrows the queue, it does not make the final call alone.
Automated security monitoring and alert triage
Security alerts from endpoints and infrastructure are triaged automatically to separate routine noise from items that need analyst attention. A person investigates and decides on any action affecting a client.
Instant quote system
Our instant quote tool walks a prospect through a common core of questions plus an industry-specific questionnaire, and produces a quote without waiting for a salesperson to build one by hand.
Responsible AI and privacy
We built Hexafusion Labs around a short list of rules that do not bend for convenience.
- Client data is never used to train AI models.
- A person reviews decisions that affect a client before those decisions take effect.
- A caller is told plainly that they are speaking with an AI.
- We never collect card numbers or passwords by phone, whether the call is answered by a person or by an AI.
- BC's Personal Information Protection Act (PIPA) and the federal Personal Information Protection and Electronic Documents Act (PIPEDA) guide how we collect, use, and store personal information in these systems.
This is general information, not legal advice.
Research notes
A few plain-language notes from reviewing recordings of our own AI phone calls. We keep updating this list as we learn.
- Tell callers early when they have reached us after hours, so they are not left wondering why no one picked up live.
- Connect a known client straight to a person during business hours rather than making them work through menus first.
- Never let the assistant accept card numbers under any circumstance.
- Do not apologize repeatedly. One clear apology, then move the call forward.
- Always confirm a callback number before the call ends, even if the caller ID looks correct.
Pilot programme for clients
If you would like to try an AI assistant for your own phones, or a Microsoft Copilot setup for your organisation, we can scope a pilot with you. Pilots are small on purpose: a limited scope, a short review period, and a plain-language report of what worked and what did not.
To start a conversation about piloting an AI assistant or a Copilot setup, visit our contact page and tell us what you have in mind.
AI phone assistant for businesses
What a fully automated phone assistant does, how we build and manage one, and what changes in your phone system.
Read moreCurious what else we are working on
Hexafusion Labs projects move into our regular services once they are proven. Reach out if you want to pilot one with us first.
Contact HexafusionRelated pages
Hexafusion Labs projects support our wider services and industry work.
