Line-of-business system agents
A custom MCP server that gives your agents governed read and write access to your ERP, finance or CRM platform, so answers come from the system of record.
Custom Build · Part of the AI Fast Start programme
Custom AI, built to your stack.
The three fixed tracks ship a working agent on one platform in weeks. Custom Build is for the work that runs wider: bespoke MCP servers, agents that reach every system that matters, and multi-agent workflows scoped to your environment. Designed, built and shipped by the team that runs the DI AI Foundry.
Atlassian Solution Partner
Melbourne HQ, nationwide service
The opportunity: connect AI to everything that matters
Most teams start with one platform and one agent, and that first win proves the value fast. The bigger step is agents that reach across your systems, and workflows that carry work from end to end.
Agents that read from your ERP, service desk, document stores and line-of-business systems
Custom MCP servers that give your agents a governed, permissioned bridge into the systems standard connectors do not reach
Multi-agent workflows that turn isolated helpers into coordinated work, with a human in the loop where it matters
The organisations that get ahead make this move deliberately, with governance and documentation built in from the first step.
One method. Ten steps. Built on the DI AI Foundry.
Every AI Fast Start engagement follows the same proven 10-step methodology. The foundation is consistent across every track. On Custom Build, the configuration is shaped around your systems, your data and the agents that need to work together.
01
Governance and security baseline
AI access policies, security documentation and audit settings established first, so every build that follows is approved faster.
02
Access and permissions modelling
Who can build, who can use, and what each agent is allowed to reach. Permission structures documented and implemented across teams, projects and data sources.
03
Use case discovery and prioritisation
Identify the workflows where a custom build delivers the most value, and sequence them. Strategic planning grounded in your environment, not generic AI hypotheticals.
04
Platform fit and stack confirmation
Confirm the platforms, systems and data sources in scope, and where custom MCP servers are needed to reach them.
05
System connection and data integration
Design and stand up the connections between your agents and your systems. This is the step where Custom Build goes further than any fixed track.
06
AI Prompt Library
Deploy a curated AI Prompt Library with pre-built, tested templates tailored to your team's most common workflows and use cases.
07
MCP server and agent design
Design the custom MCP server interfaces and the multi-agent architecture: which agents exist, what each one owns, and how they hand work to each other.
08
Agent build and deployment
Build and test the agents against real workflows, then deploy them into your environment.
09
Your track takes over
Custom MCP and multi-agent build
Build, integrate and ship the custom MCP servers and the coordinated multi-agent system, with the documentation to run and extend it.
10
Enablement, handover and 90-day roadmap
Hands-on training for your team, a documented asset register, and a 90-day roadmap for what to build next.
Which track fits your first build?
Most teams start with a fixed track and grow into a custom build. Some start custom because the highest-value workflow already spans systems. Your discovery call settles it in 30 minutes.
| Track | Best when | Investment | Timeline |
|---|---|---|---|
| Claude Track | Your team has Claude and needs capability fast | $5,000 + GST | 2 to 3 weeks |
| Rovo Track | Your work lives in Jira and Confluence | $5,000 + GST | 2 to 3 weeks |
| Copilot Track | Your organisation runs on Microsoft 365 | $6,500 + GST | 3 to 4 weeks |
| Custom Build (this page) | Your workflows span systems, need bespoke integrations, or need agents working together | From $15,000 + GST | 4 to 8 weeks |
What teams build with us
Custom is abstract until you see your own problem on the page. These are the shapes of build this track delivers.
Multi-agent service triage
Agents that classify, enrich and route incoming requests across your service desk and knowledge base, handing each other work with a human in the loop where it matters.
Cross-platform reporting
An agent that assembles status, risk and delivery reporting from data spread across project, finance and time-tracking tools.
Document and compliance workflows
Agents that read your policy and contract stores through a permissioned MCP bridge, and draft, check or summarise against your standards.
Legacy system integration
A custom MCP server in front of an internal or legacy system, so modern AI agents can work with the tools you already rely on.
Your workflow
Every organisation has processes that do not fit a standard category. If the work matters and it spans systems, it is a candidate.
A governed AI capability, not a proof of concept
Custom MCP servers
Bespoke, permissioned bridges between your agents and the tools and data the standard connectors do not reach.
A multi-agent system
Several agents with clear ownership and clean handoffs, scoped to your highest-value workflows.
Governance built in
Access policies, verification rules and security documentation established in step one, not bolted on at the end.
Integration documentation
Every interface, permission and data flow documented, so your platform and security teams can sign off with confidence.
Asset register
Agents, MCP servers, prompts and patterns documented in your own environment, so the capability is owned by you.
Enablement and roadmap
Hands-on training for your team plus a 90-day roadmap, so the build keeps compounding after handover.
How a custom engagement is scoped
01
Step 01 · Discovery
Discovery call
Free, 30 minutes. We map your systems, your highest-value workflows and your constraints, and confirm Custom Build is the right track.
02
Step 02 · Design
Design phase
A short, fixed design engagement that produces the architecture: which agents, which MCP servers, which integrations, in what order.
03
Step 03 · SOW
Fixed statement of work
You get a scoped SOW with a fixed investment and timeline before the build starts. No open-ended engagements.
04
Step 04 · Build
Build and ship
4 to 8 weeks. The build runs to the SOW with Formula One precision: systematic delivery, weekly checkpoints, working software in your environment.
Fixed scope. Clear investment. No surprises.
Every custom engagement is priced from a fixed statement of work, set after the discovery call and design phase. The investment reflects the systems involved, the number of agents, and the integration work. You know the number before the build starts.
Custom Build
Custom AI Build
From $15,000+ GST · Scoped SOW
Bespoke MCP servers for your systems of record
Multi-agent workflows, human in the loop
Governed, permissioned data access
Governance and documentation built in
From first build to compounding capability
The Custom Build ships a working system and sets a clear direction. To keep extending it, adding agents, connecting more systems, optimising licensing and improving the platform underneath, organisations move into a partnership with Design Industries' Digital Factory: our three-pillar managed service covering licensing optimisation, platform support, and strategic improvement hours. Custom development continues as strategic improvement work, so the capability keeps compounding.
Licensing optimisation, platform support, and strategic improvement hours.
Why Design Industries
We built it for ourselves first
The MCP servers, agents and multi-agent workflows we ship are the same patterns we run inside our own business every day. The lessons are real and current.
Deep MCP integration expertise
We are specialists across Anthropic's Claude, Microsoft Copilot and Atlassian Rovo, and we connect all three to the wider stack through custom MCP servers. Early adopters, with production deployments behind us.
Governance-first delivery
Security baseline and permissions modelling are steps one and two of every engagement. Built to meet enterprise and government requirements, not work around them.
Formula One precision
Every optimisation matters. Every day counts. Systematic delivery to a fixed SOW, in weeks, with no discovery paralysis and no scope creep.
Frequently asked questions
How is Custom Build different from the Claude, Rovo and Copilot tracks?
The three fixed tracks each ship a working agent on one platform, at a fixed price and timeline. Custom Build is for work that runs wider: agents that reach systems the standard connectors do not cover, custom data access through MCP, or several agents working together. It is scoped to your environment and priced from a fixed statement of work, from $15,000 plus GST.
What is an MCP server, in plain terms?
MCP (Model Context Protocol) is an open standard that lets an AI agent connect to your tools and data through a defined, permissioned interface. A custom MCP server is the bridge we build so your agents can read from and act in the systems that matter to you, with your access rules respected.
Do we need to have run a fixed track first?
No. Many teams come to Custom Build after a fast start, because the foundation steps are already in place and the build moves faster. Teams whose highest-value workflow already spans systems start here directly, and the ten-step method covers the foundation either way.
How is a custom engagement scoped and priced?
A free discovery call, then a short design phase, then a fixed statement of work. Pricing starts from $15,000 plus GST and reflects the systems involved, the number of agents, and the integration work. You know the number before the build starts.
What is the typical timeline?
Most custom engagements run 4 to 8 weeks from kickoff to a live build, depending on scope. The design phase sets the timeline in the statement of work.
Which platforms can a custom build work across?
We build across Anthropic's Claude, Microsoft Copilot and Atlassian Rovo, and we connect to your wider stack through custom MCP servers. The right mix is set by your highest-value use case, not by a default platform.
Can agents read data from outside our main platforms?
Yes. That is the core of this track. Custom MCP servers give agents governed access to line-of-business systems, document stores, databases and legacy tools, with permissions respected and every interface documented.
How are security and governance handled?
Security and governance are steps one and two of the method, not an afterthought. Every engagement starts with a security baseline and permissions modelling, and every MCP interface ships with documented access rules your security team can review.
What happens after the build?
You have a working system in your environment, a documented asset register, a trained team and a 90-day roadmap. Most organisations continue through a Digital Factory partnership, where custom development runs as strategic improvement work.
Proven in Australian environments
Tell us what you are trying to build
Book a free 30-minute consult. We will map your systems, identify the highest-value first build, and outline the pathway to a scoped statement of work. No obligation. No sales pressure. Just a straightforward conversation about what custom AI can do in your environment.
Part of the AI Fast Start programme. View all tracks.
Free 30-minute consult
A 30-minute consult to scope the build, no obligation.
We map your systems, your data and the agents that would move the needle
You leave with a clear scope, an indicative timeline and a next step
Built and run by the team that runs production AI in its own business