The human side of AI
Agents do the legwork, and your people make the calls.
Ludia Consulting is an AI-forward Frontier Firm that specializes in Service Operations excellence and architecture. AI is in the fabric of how we work. We show leaders where it pays back in growth and margin, what it costs, and where people stay in charge, then measure it in your numbers.
Where AI fits at Ludia
We use AI as one part of a well-built Service Operation.
Ludia Consulting is a Microsoft solutions integrator focused on Service Operations. On Microsoft Dynamics 365 we connect customer service, field service, mobility, supply chain, and finance into one operation. AI agents are the tools that keep work moving between the people in it, and on this page they get the spotlight.
We don’t sell AI. Our senior consultants decide with your leaders where each agent belongs in the operation and what stays human. Then we build it, and train your people to run and improve it after we step back.
How we connect a Service Operation →Map
Our consultants interview your teams and decide, row by row, where AI belongs and what stays human.
Enable
Named owners, documented designs, and trained admins, so your team runs it without us.
Four questions to answer before you fund AI
Answer these four and you have a plan a CEO can back and a CFO can fund. Helios 2 answers them for one business line at a time, with the people who own the numbers.
Return
Where does it pay back?
Start from the measure a business line is already held to, such as schedule compliance or cost to serve. Fund the work that moves that measure.
Cost
What does it cost to run?
Ask for the running cost alongside the build: licenses, usage-based Copilot Credits, and the time your people spend reviewing what agents produce.
People
What stays human?
Safety, pricing, compliance, and care stay with people, by design. Deciding that first is what makes the plan credible to your board and your workforce.
Fit
Does it need AI at all?
About four in ten items we map need no AI model. A rule-based workflow gives the same answer every time and carries no model usage to pay for.
The Microsoft AI we build on
Copilot and agents from Microsoft, put to work inside your Service Operation, with people keeping the decisions.
Agents in Microsoft Dynamics 365Microsoft’s agents for service, scheduling, procurement, and finance, put to work in your operation.on microsoft.com (opens Microsoft’s site)
Microsoft Copilot StudioCustom agents built on your manuals, jobs, and processes.on microsoft.com (opens Microsoft’s site)
Microsoft FoundryWhere we build, test, and govern custom AI agents and models.on microsoft.com (opens Microsoft’s site)
Microsoft 365 CopilotCopilot in the tools your teams already use, connected to Dynamics 365.on microsoft.com (opens Microsoft’s site)
Microsoft Agent 365One place to see, govern, and secure every agent in the company.on microsoft.com (opens Microsoft’s site)
Microsoft AzureThe secure cloud underneath: identity, integration, data, and monitoring.on microsoft.com (opens Microsoft’s site)
Microsoft FabricOne place for operational data, analytics, and the data agents learn from.on microsoft.com (opens Microsoft’s site)How we build AI: from the agents in Dynamics 365 to Microsoft Foundry
Ludia implements AI end to end, on Microsoft’s stack. We start with the simplest layer that does the job, and go deeper only when the work needs it. Helios 2 decides which layer each step gets.
1 · In the app
The agents already inside Dynamics 365
Case management, scheduling, and finance agents work inside the system your people already use. Often the fastest payback, with the least to maintain.
2 · Configured
Copilot Studio for the agents your team owns
Agents built on your manuals, jobs, and processes, configured with your admins so they can change them without us.
3 · Built
Microsoft Foundry when the work needs more
Custom agents and a choice of models for work the packaged agents don’t cover, such as reading field tickets and documents or multi-step work across systems. Built, tested, and versioned like any other software.
Grounded
Your data, with your permissions
Agents answer from 
Dataverse, Microsoft Fabric, and your own documents, and see only what the person using them is allowed to see.
Secured
Azure underneath
Identity through Microsoft Entra, integration with the systems you keep, and monitoring and cost controls in the Azure tenant your IT team already governs.
Governed
Evaluated before it touches the work
Every agent is tested against real cases before launch and monitored after it. A person approves anything a customer sees.
Agents across your Service Operation
The agents we build work in every link of the Service Operation from the first call to the paid invoice. In each link, a person stays in charge of the call that matters. Microsoft’s status is stated as of September 2026, and we tell you which agents are still in preview.
- 1 · Customer service
Customer Service (opens Microsoft’s site)
Contact Center (opens Microsoft’s site)Every request understood the first time
Microsoft’s agents: Case Management, Customer Intent, and Knowledge Management in Customer Service (generally available); the Service Operations Agent in Contact Center (preview).
Stays human: exceptions, goodwill, and the hard conversation with an upset customer.
- 2 · Field service
Field Service (opens Microsoft’s site)The dispatcher plans the whole day ahead
Microsoft’s agent: the Scheduling Operations Agent in Field Service (preview, not yet for production use).
Stays human: the dispatcher’s final schedule and every promise made to a customer.
- 3 · MobilityField Service (opens Microsoft’s site)
Power Apps (opens Microsoft’s site)A second-year technician with a veteran’s history
Knowledge agents we build in
Copilot Studio put the asset’s history, manuals, and past fixes on the technician’s phone or tablet.Stays human: the repair call and every lockout/tagout sign-off.
- 4 · Supply chain
Supply Chain Management (opens Microsoft’s site)The buyer chases fewer suppliers
Microsoft’s agent: the Procurement Agent in Supply Chain Management (preview), which handles routine supplier follow-up.
Stays human: which supplier to commit to, and at what price.
- 5 · Finance
Finance (opens Microsoft’s site)The controller reviews the reconciliations agents prepare
Microsoft’s agent: the Finance Agent for reconciliation, variance, and collections. The finished job now reaches finance through Microsoft’s connected Field Service,
Project Operations, and Finance workflows (generally available, September 2026).Stays human: pricing, write-offs, credit decisions, and signing off the close.
Across every link
Our consultants map where agents pay back with Helios 2 and build them on Dynamics 365 and Copilot Studio. Agent 365 governs them, so every agent has an identity, a named owner, and a person reviewing its work. Ludia’s own tools start with Helios 1, which runs our own firm, and Helios 2, the version we build with your leaders. The AI Implementation Methodology is our named method for planning, compliance, and long-term scale. A national facility-maintenance services provider gave its technicians a virtual assistant on Field Service, built with us in Copilot Studio and Azure OpenAI.
Underneath it all
Azure (opens Microsoft’s site)Microsoft Foundry (opens Microsoft’s site)Dataverse (opens Microsoft’s site)
Microsoft Azure is the secure cloud every link runs on: identity, integration with the systems you keep, and the data that ties them together. In Microsoft Foundry we build, test, and govern the custom AI agents and models your operation needs beyond the ones inside Dynamics 365. They work from your own data and are evaluated before they touch the work. How we build AI →
The manifesto
Picture a caseworker with an unhurried hour for every family. Picture a dispatcher who plans the whole day before it starts.
Nobody in those pictures lost their place.
That is the change coming to work: more reach for the people already doing it.
Each person can do far more than they could alone before. Researchers call it superagency.
It’s the biggest change in working life in a century. We intend to make sure it’s human-led.
AI agents are getting good at execution. They’re also unreliable in ways that don’t look obvious: strong on one task, confidently wrong on the next one that looks just as hard.
That makes the people who direct them more important. They decide what the work is for, what the agents may do, and when the answer is theirs to own.
Most workers are ready for this. Most companies still run on processes, data, and incentives built for the old way of working, and that’s where the next decade is won or lost.
Changing them is consulting work, and it is the work we do. We map where agents belong and where people must stay in the lead.
Then we build it on Microsoft and measure what changed in the business. We ran it on our own firm first.
We’re a senior, minority-owned Microsoft solutions integrator, and we want the gains to reach everyone: the frontline, the mid-market, the nonprofit down the street.
The question is how far your people will go, not how many your company will need.
What AI should do for people and the line we draw
Each pillar is a human outcome with a mechanism behind it. The stories are patterns from the work, not claims about a named client.
Work handed back
Agents take the drafting, searching, re-keying, and chasing. People get those hours back for customers and for the calls that need their judgment.
A construction project manager used to spend Friday nights reconciling cost reports. Now a draft cost-to-complete is waiting each week. Change-order negotiations stay hers.
Expertise that doesn’t retire
The knowledge in a veteran’s head reaches the newest hire on day one.
A second-year field technician facing an unfamiliar fault gets the machine’s history and likely causes on a tablet in the cab. The repair call and the safety call stay with the technician.
Decisions that stay human
Safety, pricing, compliance, and care are marked as human-owned in every map we build.
A compressor-station tech sees procedures and past fixes at the asset. Every lockout/tagout sign-off stays a person’s call, by design.
Access, widened
Capability that once needed an enterprise budget now fits a mid-market firm, a public agency, or a nonprofit.
A shelter caseworker’s grant reports are drafted from case records, so the afternoon goes back to families. Through Dynamics for Good, we bring the same senior team to nonprofits.
Measured in your own reports
Progress shows up per business line, in the reports your leaders already read.
The maintenance planner who owns schedule compliance sits in the room when the measure is agreed. Helios 2 gives that conversation as much time as the map itself.
The line we draw
Some of the most important rows on any map say “stays human.” They’re what make the rest believable.
How Helios 2 draws it →What the evidence says
The gains are real in structured work, and largest for newer workers. They don’t reach the company until the work around people is redesigned.
Sources: Brynjolfsson, Li and Raymond, Quarterly Journal of Economics (2025), 5,179 customer-support agents at one firm. McKinsey, The State of AI (2026), 1,719 respondents surveyed May to June 2026. Microsoft, Work Trend Index (May 2026), 20,000 AI users in 10 countries; a vendor survey.
The honest counterpoint
The frontier is jagged, so the work has to be mapped.
In a field experiment with 758 BCG consultants, those using AI produced work rated more than 40% higher in quality on tasks inside AI’s frontier. On a task outside it, they were 19 percentage points less likely to get the answer right.
The design response is the map. Decide which side of the line each task sits on, and keep a person reviewing what agents produce.
Dell’Acqua et al., Organization Science (2026), GPT-4 in 2023. One more caveat: there are no randomized studies yet of field technicians or plant crews. Our frontline stories lean on support-desk evidence, and we say so.
Helios 1 → Helios 2
We mapped our own firm before we offered the map to anyone.
Helios started as how we run Ludia. We needed to know where AI should sit in our own business. So we mapped four service lines and 34 functions and found 150 opportunities.
Each one is scored, sequenced, owned by a named person, and marked as an agent, a workflow, or work that stays human. That’s Helios 1.
Customers saw it and asked for their own. Helios 2 is the same map for your business, built with your own people. You keep it whether or not we build anything next.
How Helios 2 works →How we work with AI responsibly
A named owner for every agent
Every agent we build has a named person on your side who owns it, and a person reviewing what it produces.
Human-owned from the start
Work that involves safety, pricing, compliance, or care is mapped as a process people own before anything is built.
Your data stays under your controls
We build in your Microsoft tenant, under your security, identity, and compliance controls. Microsoft\’s own Product Terms and Data Protection Addendum set how Microsoft handles your data.
We tell you when a task doesn’t need AI. About four in ten don’t.
For your IT team
What your IT team will ask about Microsoft’s AI
We build on Microsoft because your people already work there. Two layers matter most: one gives agents context, the other keeps them under control.
Intelligence layer
Work IQ
Grounds Microsoft Copilot and your agents in how your company works: Microsoft 365, Dynamics 365, and Power Platform data. Its APIs became generally available in June 2026.
Trust layer
Agent 365
The control plane to see, govern, and secure every agent. Generally available since May 2026, and included in Microsoft 365 E7. Entra Agent ID gives each agent its own identity.
Where the work runs
Dynamics 365 agents
Some are generally available, such as finance reconciliation in Microsoft 365 Copilot. Others, like the Field Service Scheduling Operations Agent, are still in preview. We tell you which is which.
Your IT team uses Agent 365 to set what each agent is allowed to do.
The Superagency Series
The Superagency Series
Mapping the jagged frontier: how to decide which work stays human
A test for each task, and why the rows marked “stays human” are the ones that earn trust.
The Superagency Series
AI helps new hires most when veterans’ fixes are on record
What support-desk research says about newer workers and AI, and what it means for the trades.
The Superagency Series
The return on AI depends on operating-model work nobody budgets for
Why organizational factors decide most of AI’s impact, and where to start.