Fractz

What we do

Conversational CX

Chatbots and voice assistants that convert and support 24/7.

Talk to an engineer

What we do

Assistants on the channels your customers already use, scoped tightly enough to be safe. The engineering that matters is not the conversation — it is the boundary: what the assistant handles, and how cleanly it hands over the moment a case falls outside it.

  • Explicit boundary and clean escalation
  • Article 50 disclosure built in
  • Answers grounded in your own content

You probably need this if

What we build

01

A defined boundary

Written before anything is built: what the assistant answers, what it refuses, and what it escalates. This is what makes automated support safe in a service business.

02

Answers grounded in your content

Retrieval over your actual policies, prices and documentation, with refusal outside that scope instead of an improvised answer.

03

Channel and system integration

Web, WhatsApp and Messenger on the front, CRM, PMS or booking system behind — so the assistant can act, not just advise.

04

Disclosure and transcript handling

AI interaction disclosed in the interface as Article 50 requires, with transcript retention set to a period you choose and can defend.

Typical stack

  • LangGraph
  • OpenAI
  • Anthropic
  • vector search
  • WhatsApp Business API
  • Twilio
  • your CRM / PMS

How an engagement runs

  1. Weeks 1–2

    Boundary and content audit

    Intent scope, escalation rules, and an honest look at whether your documentation can support good answers.

  2. Weeks 3–5

    Build and supervised pilot

    Live on one channel with every conversation reviewed, tightening the boundary against real traffic.

  3. Week 6+

    Rollout

    Additional channels and languages, with quality monitoring and a monthly review of escalations.

Process

  1. 01

    Discovery & Audit

    Goal analysis, process mapping, and KPIs.

  2. 02

    Tailored build

    Rapid development with biweekly sprints.

  3. 03

    Deploy & Scale

    Monitoring, training, and continuous optimization.

Engagements

  1. 01

    Fashion manufacturing

    Product operations for a leather goods maker

    Pellemoda S.r.l.

    A production platform the marketing team runs without engineering support

    System
    Digital product platform + operations automation
    Duration
    Multi-phase
    Read the engagement
  2. 02

    Education

    Marketing automation for an international college

    H-Farm College

    Manual marketing tasks automated; team time moved back to campaigns

    System
    Marketing automation + reporting
    Duration
    Multi-phase
    Read the engagement

Common questions

Will it make things up?
It is constrained to your content and instructed to refuse outside its boundary, and we test that behaviour with adversarial cases before launch. The escalation path exists precisely because no assistant should improvise on an edge case.
Do we have to tell customers it is AI?
Yes. EU AI Act Article 50 transparency obligations have applied since 2 August 2026: people must be told when they are interacting with an AI system. We build the disclosure into the interface.
What happens to conversation transcripts?
They are personal data. Retention period, access and deletion are set by you and implemented so an erasure request actually reaches them.
How do you measure whether it worked?
Deflection rate and escalation quality, measured against the baseline volume before launch — not conversation counts, which only measure traffic.

Services