BIM Model to Cost Estimate Integration
Metwaves built an integration that reads IFC building models and loads components and quantities into the estimating and ERP systems, cutting quantity take-off from days to hours.
Learn More >
Goal
Give staff instant, reliable answers to everyday IT, HR and policy questions inside Microsoft Teams, so the internal teams spend their time on requests that genuinely need a person.
Solution
A Copilot Studio agent in Teams that answers from the firm's SharePoint content and an Azure AI Search index, cites the policy it used, and opens a ticket with the conversation attached when it cannot help.
Result
Between 30% and 50% of staff requests are now resolved without a human. The agent went live in 8 to 12 weeks.
Technologies used: Microsoft Copilot Studio, Microsoft Teams, SharePoint, Azure OpenAI, Azure AI Search
Background
Our client is a professional services firm in the United Kingdom. Like most firms of its kind, its IT and HR teams answered the same questions every week: how to request leave, how expenses work, how to reset access, what to do when the VPN will not connect.
The answers already existed, spread across SharePoint sites and policy documents that staff found hard to search. The firm wanted an assistant inside Microsoft Teams, where its people already work, that would answer correctly or get out of the way quickly.
Challenge
A helpdesk agent is only useful if staff trust it. A confident wrong answer about a leave or expenses policy does more damage than no answer at all.
1
Scattered, overlapping content
Policies lived in several SharePoint sites, with older versions alongside current ones.
2
Answers people can check
Staff needed to see which document an answer came from, so they could rely on it.
3
Respecting permissions
Some documents are restricted to certain groups. The agent must never show content a person cannot open.
4
A clean hand-off
When the agent cannot help, the service desk must receive the request with context, not a blank ticket.
Our Approach
We started from the questions staff actually asked, not from the documents, and measured the agent against those questions before anyone outside the pilot group used it.
01
Question review
We grouped a sample of past IT and HR requests by topic to find the questions worth automating first.
02
Content clean-up
With the firm's content owners we marked the current version of each policy and retired duplicates.
03
Agent design
We designed the Copilot Studio topics, the knowledge sources and the rules for when to hand off.
04
Search index
An Azure AI Search index over the policy library, with Azure OpenAI generating grounded answers that cite their source.
05
Pilot
A pilot group used the agent in Teams while we tracked answer quality and the questions it could not handle.
06
Rollout and monitoring
Firm-wide rollout, with analytics on resolution and unanswered questions feeding the content backlog.
Answers staff can check
Every answer is grounded in the firm's own documents and shows its source, so staff can open the policy rather than take the agent's word for it.
- Knowledge sources: SharePoint sites connected directly, plus an Azure AI Search index for the policy library.
- Grounded answers: Azure OpenAI writes the answer only from retrieved passages and cites the document it used.
- Current versions only: Superseded policies are excluded from the index so an old rule cannot resurface.
Hand-offs that do not start from zero
When a request needs a person, such as a hardware fault or an exception to policy, the agent says so and passes it on.
- Ticket with context: The agent opens a ticket in the firm's service desk with the conversation and category attached.
- No dead ends: Staff can ask for a person at any point, and the agent hands over instead of looping.
- Permissions respected: Answers only draw on documents the person asking is allowed to open.
Measuring what the agent resolves
We agreed the measures with the firm before launch and report them each month.
- Resolution rate: The share of conversations that end without a ticket, now between 30% and 50%.
- Unanswered questions: Questions the agent could not answer become a backlog of content to write or fix.
- Hand-off reasons: Tickets created by the agent are tagged, so the service desk can see why they arrived.
The Results
30–50% resolved without a human
Between 30% and 50% of staff requests are answered by the agent end to end, without a ticket.
Answers inside Teams
Staff get policy answers in the tool they already use, with the source document one click away.
Better tickets
Requests that do reach the service desk arrive with the conversation and category attached.
Live in 8 to 12 weeks
From the first question review to firm-wide rollout.
Client name withheld. Details that could identify the firm have been left out.
AI Invoice Processing Agent for SAP Business One
Metwaves built a LangGraph agent that reads around 1,000 supplier documents a month with Azure Document Intelligence and posts them to SAP Business One, cutting manual data entry by about 80%.
Learn More >
AI Material Recommendation System for BIM
Metwaves built and deployed an AI recommendation engine inside BioBuild's BIM workflow that scores materials on cost, carbon and structural fit. Live since January 2025: selection time down 40%, material cost down 20%.
Learn More >INSIGHTS & NEWS
Discover IT Industry Insights and Technology Trends
Nikit Nandan
November 12, 2026
Top AI agent development companies for small and mid-sized businesses (2026)
Nine AI agent development firms a small or mid-sized business can shortlist, with Clutch figures, minimum budgets and the weaknesses their own clients name. Published by Metwaves, which is not ranked.
Learn More >
Nikit Nandan
November 12, 2026
Copilot Studio vs ServiceNow vs SAP Joule vs Agentforce: An Architecture Review (2026)
An architecture review of the four big enterprise agent platforms: where each one reasons, acts, finds its data and connects to outside agents, what it costs, and when a custom build is the better choice.
Learn More >
Nikit Nandan
November 01, 2026
Top custom software development companies for small and mid-sized businesses (2026)
Ten custom software firms a small or mid-sized business can shortlist, with Clutch figures, minimum budgets and the weaknesses their own clients name. Published by Metwaves, which is not ranked.
Learn More >Let’s build your idea!
Let us help you with free analysis & estimate.
Our product expert will contact you within 12 hours.