First legal app certified “Works with Microsoft 365 Copilot

Matter-centric documents: LawToolBox × iManage

Keeping deadlines and documents in lockstep — automated filing, matter sync, and a single source of truth across both systems.

The Challenge

Copilot was built for the enterprise. Legal was an afterthought.

Three problems stood between a generic AI assistant and a legal-grade copilot.

No legal context in the base product

Ask Copilot for “upcoming deadlines” without a legal agent and you get birthdays and dentist appointments mixed with filing dates.

Certification is a gauntlet

Microsoft 365 App Certification requires security controls, evidence collection, publisher attestation and a 14-day initial document submission — before a 60-day full review.

Privilege cannot be renegotiated

Legal AI must never leak client-matter data. Every packaging, permission and grounding decision had to satisfy privilege requirements first.

The Path

From application to approval — the certification gauntlet

Microsoft’s certification framework is not a checkbox. It is a multi-phase evidence process.

Phase 1 · 14 days

Initial documentation submission

ISVs submit architecture overviews, support environment documentation and security control mappings. This is the scoping phase — get it wrong and the whole timeline slips.

Phase 2 · 60 days

Full evidence review

Upload evidence against every applicable control defined in scope. Security, data handling, authentication, and compliance artefacts all get scrutinised.

Publisher Attestation

Partner Center completion

Complete the Publisher Attestation form in Partner Center, aligning with Microsoft Commercial Marketplace certification policies.

 

December 11, 2023

Approved: “Works with Copilot”

LawToolBox becomes the first legal application approved for Microsoft Teams Copilot — at a time when Copilot for Microsoft 365 was only available to enterprises purchasing 300+ licences.

 

Architecture

How the Copilot agent works

A four-stage flow from matter data to grounded Copilot response — all inside the customer’s Microsoft 365 tenant.

Data flow · LawToolBox Copilot agent on Microsoft 365

Matter Containers

M365 groups · SharePoint sites · Teams · calendars provisioned per matter

LawToolBox Agent

Reads matter data · deadlines · documents · saved AI insights

Copilot Grounding

Queries resolve against matter containers — not the whole tenant

Teams · Outlook · Copilot

Conversational queries · deadline reports · action from chat

Figure 1. The agent sits between Microsoft 365 matter containers and Copilot surfaces, ensuring every query resolves against client-matter data.

Figure 1. The agent sits between Microsoft 365 matter containers and Copilot surfaces, ensuring every query resolves against client-matter data.

86 deadlines auto-generated for a single matter, each with trigger date, rule citation and source document.

86 deadlines auto-generated for a single matter, each with trigger date, rule citation and source document.

What We Built

Four capabilities that made Copilot legal-grade.

Matter-grounded Copilot answers

The LawToolBox agent anchors every Copilot query to client-matter containers. Ask for “upcoming deadlines” and you get a clean, legal-only list — not a mixed calendar.

Community prompt library

An out-of-the-box legal prompt library, plus firm-level customisation. Designated prompt experts tailor community prompts to practice areas and expertise.

AI-ready matter containers

We provision SharePoint sites, Teams, Group Calendars and Outlook entries per matter. Copilot inherits Purview sensitivity labels and resolves queries against the right container.

Copilot-native, not Copilot-adjacent

Every time Microsoft expands Copilot into a new surface — Excel, PowerPoint, Teams meetings — LawToolBox is automatically there. No separate build, no lag.

86 deadlines auto-generated for a single matter, each with trigger date, rule citation and source document.

86 deadlines auto-generated for a single matter, each with trigger date, rule citation and source document.

Inside the build

Four engineering decisions that got us through certification — and kept privilege intact.

Certification strategy

Scope first, build second

We defined the certification scope before writing a line of integration code. Every control was mapped to an owner, an evidence source and a deadline in the 60-day review window.

Grounding layer

Containers, not indexes

Instead of a flat search index, we provision per-matter M365 containers. Copilot queries resolve against the container, not the tenant — which is what makes answers legally relevant.

Security model

Purview labels inherited

Client matters inherit Purview sensitivity labels automatically. No separate security model — it rides on Microsoft’s own compliance infrastructure.

Ecosystem leverage

Ride every Copilot expansion

Because we built natively on Microsoft’s agent framework, every Copilot surface expansion carries the legal capability automatically. Zero incremental engineering per surface.

Security is not a feature. It is the architecture.

Legal AI lives or dies on trust. Every design decision started with privilege, residency and auditability.

Data stays in tenant

Matter containers live inside the customer’s Microsoft 365 boundary. No data leaves for model training.

MACC eligible

Aligns with Microsoft Azure Consumption Commitment, making procurement straightforward for enterprise legal departments.

HIPAA-configurable

Built on Microsoft Azure Cloud, configurable for HIPAA compliance where legal matters intersect with healthcare data.

Impact

What changed after Copilot certification.

1st

Legal application certified as “Works with Copilot” for Microsoft Teams

86

Deadlines auto-generated for a single matter — each with rule citation and source document

Global

Deployed in Microsoft Innovation Hub demo environments for field sellers and architects

" Our ability to access a wide range of pre-built and customisable AI models and easy-to-use APIs through Azure OpenAI helps our solution maintain the optimal level of intelligence to recognize matter-specific dates and other types of content — even from handwritten notes
Carol-Lynn Grow
COO, LawToolBox

Related work

Ready to launch your AI services offer?

Join the AI Services Partnership Program and start turning Microsoft AI demand into real business growth.