AI SYSTEMS · AUTOMATION · SOFTWARE

Systems that
do the work.

DueClix builds AI agents, automation and custom software that run real business operations — not demonstrations of them.

Start a projectExplore our work
Agents
that take actions
Automation
across your stack
Software
built to be owned

01 — Input

  • Website form
  • Inbound email
  • Missed call

02 — Reasoning

  • Qualification agent

03 — Action

  • Enrich & score
  • Draft the reply
  • Route to an owner

04 — Outcome

  • Business state changed

Qualified lead in the CRM, owner notified, reply sent

intake.received source=web_form id=8842_

The shift

Software is no longer something a business uses. It is how the business runs.

Yesterday

Software that records the work

Forms, dashboards and databases that store what a person already did. Useful, but every outcome still waits on someone with the time to produce it.

What we build

Software that does the work

Systems that read what arrives, decide what it means against your rules, act across the tools you already run, and leave a record of every step. The outcome stops being someone's task.

Try it

What should we automate?

Describe something your business still does by hand. We'll map it into the system that would do it instead.

Press ⌘+Enter to map it.

Or start from one of these

Your process becomes a five-stage system: what triggers it, what it understands, what it decides, what it does, and what changes as a result.

What we build

From AI ideas to systems that actually run.

Six things we build. Every one of them ends in something your business can operate, extend and own.

01

AI agents

Software that reads context, decides against your rules and takes the action — then reports what it did and why.

  • Tool use
  • Policy guardrails
  • Human handoff
  • Full audit trail
02

Business automation

The repetitive path between your systems, engineered once so it stops consuming people.

  • Event-driven
  • Retry-safe
  • Observable
03

Custom software

Internal platforms, portals and operational tools built around how your business actually works.

  • Owned outright
  • Documented
  • Built to hand over
04

SaaS products

Multi-tenant products from architecture through deployment — billing, roles and scale designed in from day one.

  • Tenancy model
  • Billing
  • Scales with usage
05

AI integration

Intelligence added to the systems you already depend on, without a migration project attached.

  • Existing APIs
  • Retrieval over your data
  • Incremental rollout
06

Digital transformation

Manual processes replaced with systems, in an order that keeps the business running throughout.

  • Process mapping
  • Staged cutover
  • Measured before and after
AI agents

Agents that don't just answer. They act.

A chatbot returns text and stops. An agent is given tools, permissions and a definition of done — and it operates your systems until the job is finished.

How our agents are built
InboxDocumentsDatabaseCRM / ERPChat & alertsReportingAGENT

read.source type=email attachments=2_

How we work

From idea to intelligent system.

Five stages. Each one ends in something you can look at, not a status update.

  1. 01

    Discover

    Understand the business problem.

    We sit with the people doing the work and map what actually happens — including the parts nobody documented. Most of what gets automated later is found here.

    You get: A written process map and the honest constraints.

  2. 02

    Design

    Design the experience and the architecture together.

    Interface and system are designed in the same pass, because in an AI product the boundary between them is where things go wrong. We decide what the machine does and where a human stays in the loop.

    You get: Interface direction, system architecture, decision boundaries.

  3. 03

    Build

    Develop the software and the AI layer.

    Built in working increments you can use, not milestones you receive. Evaluation is set up alongside the model work — an agent without a way to measure it is a liability.

    You get: Running software, tests, and evaluations you can read.

  4. 04

    Automate

    Connect the workflows, APIs and systems of record.

    The system is wired into the tools you already run. Failures are designed for first: retries, dead letters, and a human queue for anything the system should not decide alone.

    You get: Live integrations with monitoring and a rollback path.

  5. 05

    Scale

    Monitor, optimise and extend.

    Once it runs, the questions change: what is it costing, where is it wrong, what should it handle next. We instrument for those questions before you have to ask them.

    You get: Dashboards, cost controls, and a roadmap of the next processes.

Reference systems

The systems we build, in detail.

Four architectures we build repeatedly. Named client case studies are published only once the client approves them — so these describe the work, not a customer list.

All reference systems
AI agent

Inbound qualification system

Challenge
Enquiries arrive across a form, a shared inbox and a phone line. Whoever is free reads them, guesses at priority, and the rest wait. Nothing is scored the same way twice.
System
An agent captures every channel into one intake, enriches the company from public sources, scores fit and intent against written criteria, creates the CRM record, assigns an owner and drafts the first reply for approval.
What changes
Sales opens a ranked, enriched list instead of an inbox, and every enquiry is answered — including the ones that used to fall through.
  • LLM agent
  • CRM API
  • Enrichment
  • Queue
  • Audit log
Architecture

Five layers, and what each one owes the next.

We don't pick technology because it's current. Each layer exists to hand something specific to the one below it — open a layer to see what.

    • Next.js
    • React
    • TypeScript
    • Tailwind
    • Design systems

    Hands down: User intent and context

    • Node.js
    • Python
    • REST & GraphQL
    • Auth & RBAC
    • Background jobs

    Hands down: Validated, authorised requests

    • LLMs
    • Agent frameworks
    • Tool calling
    • Evaluation
    • Guardrails

    Hands down: Decisions with their reasoning attached

    • PostgreSQL
    • Vector databases
    • Embeddings
    • Caching
    • Pipelines

    Hands down: Facts the system can cite

    • AWS
    • Containers
    • CI/CD
    • Observability
    • Secrets management

    Hands down: Uptime, cost and a rollback path

Why DueClix

Four commitments you can hold us to.

01

Built around outcomes

We don't build technology because it's possible. If a project can't name the thing that will be measurably different afterwards, it isn't ready to start.

02

AI-native thinking

Intelligence is part of the architecture, decided alongside the data model and the permissions — not a feature added once the software is finished.

03

Automation first

Before we build a screen for a task, we ask whether the task needs a person at all. The best interface is often the one nobody has to open.

04

Built to be handed over

Documented, tested and readable, because the system has to outlive the engagement. You should be able to take it elsewhere — that's the standard we build to.

What should your business stop doing manually?

Tell us what is slowing you down. We'll tell you honestly whether it's a system worth building — and what it would take.

Start a projectMap a process first

AN ENGINEER READS EVERY ENQUIRY — NOT A SALES QUEUE