INTELLOX.AI · AI ENGINEERING & GOVERNANCE SCROLL TO EXPLORE ↓
01 // PRACTICES

Five practices, one accountable team.

Systems we deliver include monitoring, documentation and a handover plan, so your team can run them after we leave.

/010

AI agents & autonomous workflows

LangGraph · Claude · OpenAI · Python

Agents that read documents, query business systems and draft actions. Checks around the model are deterministic, and people approve high-stakes steps before anything changes.

  • Retrieval-augmented generation (RAG) over your documents and databases
  • Input and output checks, including personal-data handling
  • Tool integrations with your CRM, ERP, helpdesk and chat tools
/020

Process automation

Power Automate · UiPath · Python

Automation for handoffs between finance, operations and support, so people stop copying data between systems. Exceptions go to a review queue, not into silence.

  • Document extraction with validation rules
  • Retries that do not duplicate actions, with failure queues
  • A run log for every automated step
/030

Data platforms & Microsoft Fabric

Azure Data · Microsoft Fabric · Delta Lake · Python

We move fragmented data into governed lakehouse layers, with quality checks and lineage you can inspect. Reports and AI systems read from the same checked tables.

  • Medallion (bronze, silver, gold) lakehouse design on Microsoft Fabric
  • Data quality tests, schema checks and lineage
  • Cloud region agreed with you at the design stage
/040

Custom AI apps & product engineering

Next.js · React · Node.js · Python · Flutter

From internal operator consoles to customer-facing products, we build responsive web and mobile applications with typed APIs, automated tests and CI/CD.

  • Streaming interfaces for AI features
  • Modular services and API design
  • Documentation and onboarding for your team
/050

AI governance, EU AI Act & ISO/IEC 42001

EU AI Act · ISO/IEC 42001 · ISO/IEC 27001 · SOC 2 criteria

We classify each AI system under the EU AI Act, identify the obligations that apply to your role, and build the evidence into your engineering process: risk records, technical documentation, logs and oversight steps.

  • EU AI Act risk classification and role mapping (provider or deployer)
  • Article 50 transparency and machine-readable marking for generated content
  • Control mapping across ISO/IEC 42001, ISO/IEC 27001 and SOC 2 criteria
02 // ENGAGEMENT TYPES

The kinds of systems we design.

See case studies →
AI AGENTS & TRIAGE SCOPED PER PROJECT

Support triage agent with human sign-off

The build: a multi-step agent that drafts answers to routine tickets from approved sources, and sends anything involving money or account changes to a human approver.

THE STACK: PYTHON · LANGGRAPH · POSTGRES · AUDIT LOGS
DATA & FABRIC SCOPED PER PROJECT

Microsoft Fabric lakehouse with lineage

The build: several operational sources brought into bronze, silver and gold layers, with quality tests, personal-data masking in curated layers, and lineage records for reporting and AI datasets.

THE STACK: AZURE DATA · MICROSOFT FABRIC · DELTA LAKE
GOVERNANCE IN THE BUILD SCOPED PER PROJECT

Logging and documentation pipeline for an AI system

The build: an inference gateway that keeps event logs for at least six months where the Act applies, marks generated content, and collects technical documentation evidence on each release for a person to review.

THE STACK: EU AI ACT · ISO/IEC 42001 · CI/CD GATES

Engagement types describe patterns we design for. Each project is scoped separately.

03 // HOW WE WORK

Structured delivery, no black boxes.

We work in two-week sprints, with a working demonstration at the end of each and documentation at every stage.

STAGE 01

Diagnose

Review current workflows, data readiness, bottlenecks and regulatory exposure, including your EU AI Act role and risk class.

OUTPUT: ARCHITECTURE & RISK BRIEF
STAGE 02

Design

Define data flows, model choices, guardrails, human-oversight points and milestones for your approval.

OUTPUT: SYSTEM DESIGN & CONTROLS
STAGE 03

Deliver

Build the system inside your cloud environment with scoped access for our team, with automated tests, evaluation checks and security monitoring.

OUTPUT: WORKING SYSTEM & TESTS
STAGE 04

Operate

Handover of repositories, runbooks and training as set out in each contract, with optional ongoing support for monitoring and compliance evidence.

OUTPUT: HANDOVER PACK (PER CONTRACT) & OPTIONAL SUPPORT
04 // ENGAGEMENT MODELS

Three ways to work with us.

Fixed-scope reviews and builds, or an ongoing partnership for teams that ship continuously.

01 · 1–2 WEEKS

Readiness review

For teams scoping an AI initiative, a data migration, or a first EU AI Act and ISO/IEC 42001 gap review.

  • System and data bottleneck map
  • EU AI Act role and risk classification, with a control gap list
  • Prioritised roadmap with effort estimates
02 · SCOPED PER PROJECT

Fixed-scope system build

Milestone-based engineering to take an AI agent, data platform or governance pipeline into production.

  • Implementation inside your cloud account, with scoped access
  • Automated tests, evaluation checks and audit logging
  • Documentation, runbooks and team training
03 · ONGOING PARTNERSHIP

Embedded engineering and governance partner

Continuing delivery, model evaluation and compliance evidence for teams that are scaling AI.

  • Regular evidence updates for EU AI Act, ISO/IEC 42001 and SOC 2 criteria
  • Quarterly architecture reviews
05 // FREQUENTLY ASKED QUESTIONS

Clarity before we write a line of code.

We treat governance as part of the design, not a review at the end. Risk classification, personal-data handling, transparency measures and logging are built into the engineering process, so the evidence you need for the EU AI Act and ISO/IEC 42001 is created as the system is built.
Not always. Many projects start with data spread across ERP systems, SharePoint, Postgres and SaaS tools. We build the ingestion and quality steps your first use case needs, alongside the AI application, and extend the platform as the work grows.
Prohibitions and AI literacy measures have applied since 2 February 2025. Rules for general-purpose AI models, and the penalty provisions, have applied since 2 August 2025. Article 50 transparency obligations apply from 2 August 2026. On 2 December 2026, new prohibitions apply, and the Article 50(2) marking transition ends for generative systems placed on the market before 2 August 2026; systems placed on the market after that date must be marked from the start. Annex III high-risk obligations apply from 2 December 2027, and Annex I product-embedded obligations from 2 August 2028. These dates reflect the AI Act as amended by the Digital Omnibus on AI, as shown on the European Commission's AI Act Service Desk. Last reviewed 17 November 2026*.
We build inside your own cloud account and repositories, with access limited to what the project needs. Depending on your environment, we work in Azure, AWS or Google Cloud, or in infrastructure hosted in the EU that you already use. Ownership of code, infrastructure templates, evaluation harnesses and documentation is set out in each contract, and materials are handed over to your team. Licence terms for each project are agreed in its contract, and third-party components remain under their own licences.
No. We help clients prepare evidence against these frameworks. We do not provide certification, assurance, or legal opinions.

* Indicative, general information only; not legal advice.