Auralis vs building in-house on the Microsoft stack

Jude Rosario, Co-Founder & CEO at Auralis AI

Jude Rosario

October 30, 2026 · 5 min read

A laptop showing code on a desk
On this page
  1. What the Microsoft stack provides
  2. The e-automate connection
  3. Ownership and maintenance
  4. What it costs
  5. Where a build makes sense
  6. What Auralis does with the same jobs
  7. Four questions for your IT lead
  8. Sources

A company that already pays for Microsoft 365 may reasonably ask whether its help desk could run on top of it. Microsoft provides several components that could be used in a prototype. This comparison examines what running one in production alongside e-automate would require.

What the Microsoft stack provides

Microsoft’s help pages describe several building blocks a prototype could use. A Power Automate automated cloud flow can be triggered by the arrival of an email from a specific person. The Outlook connector can work with a shared mailbox, provided the account named in the trigger is a member of that mailbox. Copilot Studio can use SharePoint as a knowledge source, and it describes agents that answer questions and resolve issues for customers over voice and chat.

Those components cover email triggers, shared mailboxes, knowledge sources and customer conversations. They could form part of a prototype for one narrow request type, and that’s a useful starting point. Production use also needs decisions about integration access, ownership, exception handling and upkeep.

The e-automate connection

A prototype that reads from or writes to e-automate needs an approved integration path. Microsoft’s environment documentation says an app is only permitted to connect to the data sources deployed in the same environment, including connections, gateways and Dataverse databases. The team needs to work out how that restriction applies to its proposed e-automate connection, and it’s worth doing that early.

On the e-automate side, ECI describes the Equipment Solutions Network (ESN) as a data portal that vendor integrations can use to interact with its ERP software. It also says vendors are limited to a specific set of APIs per dealer integration. Before planning a build, confirm with ECI which APIs a dealer-built tool could be authorized for. Microsoft’s connector reference is the place to check whether a suitable e-automate connector exists.

Ownership and maintenance

Microsoft’s help pages also cover what happens when an automation’s owner leaves an organization. An orphaned flow no longer has a valid owner, and it can fail when it uses connections tied to that person’s account. The team should decide how ownership and connections will be reassigned if the owner leaves.

The plan should also settle who keeps the prompts, knowledge sources, exception rules and later changes up to date. A prototype might not reveal the ongoing work needed to keep these elements running, so ownership should be assigned before launch.

What it costs

Microsoft prices Power Automate per user and per bot each month. Copilot Studio includes flexible, usage-based billing, and an Azure subscription is required to use agents. The budget should also account for internal labor to build and maintain the system, because licenses alone don’t cover it.

Where a build makes sense

An in-house build may suit a narrow, stable use case: one shared inbox, one routing rule, no writes into e-automate, and an IT lead who wants to own the result. For a dealer with that requirement, keeping such a build may be reasonable.

Once a prototype works, the next request is easy to imagine, because the service manager may ask whether the tool can also handle toner, open a service call or call the customer back, and each of those would expand the integration and upkeep it needs. Reassess ownership and staffing whenever the scope expands, and don’t wait for the first failure to do it.

What Auralis does with the same jobs

Auralis handles meter reads, toner and service calls, and dispatch logging around the clock, with exceptions routed to your people. The work is written into e-automate through its ECI partner integration via the ESN API, and the service is fully managed. A meter read inside the variance you set posts on its own. A read that doesn’t meet the configured rule stops and goes to one of your people with the reason attached. Auralis runs under a service account you create, scoped to the objects you approve. Telephony covers full inbound and outbound calls. Auralis has completed SOC 2 and ISO work, and customer data is never used for model training.

One dealer we work with saw volume rise 30 to 40% after folding two acquisitions into one team. About 80% of its meter reads were automated within two weeks, and customer wait time fell from a few hours down to a couple of minutes. The same team handled that work right after the acquisitions.

Four questions for your IT lead

  1. Which APIs could ECI authorize for a dealer-built tool?
  2. Who owns the build in month six, and who covers when that person is away?
  3. How will meter-read exceptions be defined and handled, and who decides?
  4. What will the security review need to see before it signs off?

Book a 30-minute walkthrough with your IT lead on access, hosting and the audit log.

Sources

Share this

For e-automate dealers

See Auralis on your own service desk

A 30-minute walkthrough on your workflow — meter reads, dispatch, supplies and escalation.

Book a demo
Jude Rosario, Co-Founder & CEO at Auralis AI

Jude Rosario

Co-Founder & CEO

Jude founded Auralis after years as a software solutions architect at Toptal, where he saw that most support problems were really systems problems: fragmented tools, buried knowledge, and processes that broke the moment volume spiked. As CEO he sets the vision for agentic AI customer support — from the first chatbot to today's Auralis platform — and still evaluates every product decision like an architect: if it doesn't scale, it doesn't ship. He leads Auralis with a simple conviction: great support is a customer walking away feeling taken care of, and AI should be judged on resolutions, not deflections. He writes regularly on where agentic AI and the support industry are headed.

We use cookies to run this site and, with your consent, to analyze usage and improve our marketing. You can accept all, reject non-essential, or choose. See our Cookie Policy.