By Introhive
Drawing on ILTA’s 2026 Technology Survey and Introhive’s 2026 Partner Survey Report
Most professional services firms didn’t plan their AI stack so much as accumulate it. A drafting assistant for one practice, a research tool for another, an enterprise assistant rolled out firm-wide, and a few pilots the BD team started on its own. Each made sense at the time. Together, they leave your firm paying for overlapping licenses and connecting the same systems again and again.
That spend is now under scrutiny. ILTA’s 2026 Technology Survey found that, for the first time, an emerging technology has become the primary focus of law firm technology investment. Cost and ROI now rank among firms’ top three concerns about generative AI.
Adding another tool won’t fix that. When Introhive polled professional services leaders at a recent Solutions Spotlight session, only 29% said they had standardized on an AI tool and connected it to their own data. The return on those investments depends less on which tools your firm licenses than on whether each of them can draw on the same governed layer of firm context, and a Model Context Protocol enterprise architecture is an approach more IT leaders are now evaluating to build that layer without the need for another integration project.
Table of contents
Combating tool sprawl and the hidden costs of AI
The first wave of generative AI rewarded speed, and the result in most firms is AI tool sprawl. LawSites reports that 94% of law firms in ILTA’s survey are using or exploring generative AI, and many run several tools side by side: Claude, CoCounsel, and Harvey each appear at more than four in ten firms. Every one of those tools needs its own license, its own governance review, and its own path to firm data.
Accounting firms are feeling the same pressure. In the 2026 Intuit QuickBooks Accountant Technology Survey, reported by CPA Practice Advisor, 77% of accounting professionals agree the gap is widening between firms where AI is embedded and firms where it’s used only occasionally.
The licensing bill is the visible cost. The harder cost is that each tool sees only a slice of the firm. A partner preparing for a pitch can ask a general-purpose assistant to summarize the target’s annual report, but not who at the firm already knows its CFO. That answer sits in email, calendars, and the CRM, out of the assistant’s reach.
Introhive’s 2026 Partner Survey Report shows how common that gap is. Among partners using AI for business development, 68% rely only on general-purpose tools such as ChatGPT, Copilot, or Gemini. On their own, those tools have no view of the firm’s relationship network or the relationship capital it holds. Partners rate their satisfaction with AI for BD at 6.7 out of 10.
The power of composability: unifying the tech stack
If you’re building a legal tech data lake or a broader firm-wide data strategy, every AI line item is going to face the same question when it comes to budget: what revenue does this support? Your BD team is a good place to start. They can point to the specific points where partners lose time or miss opportunities, which gives you a revenue story to take to budget. In our survey, the biggest source of BD friction partners named was knowing where to focus their time (50%).
Partners often get the blame for that, but knowing where to focus is difficult when a client’s last email sits in one partner’s inbox, its billing history in the finance system, and its open matters in the practice management system, with no single tool able to see all three. A standalone AI model inherits the same blind spot however capable it is, because it can only reason about the systems it has been connected to.
Composable AI legal tech is designed to close that gap by treating each system as a building block that any approved AI tool can draw on, rather than locking each capability inside its own portal. With a Model Context Protocol enterprise architecture, your AI tools can draw on Introhive’s relationship intelligence alongside the other systems your firm connects:
- Your CRM, for pipeline, opportunities, and account plans
- Time and billing systems, for revenue history and active matters by client
- Microsoft 365, for documents and day-to-day work context
- Third-party data your firm already licenses, such as company and market intelligence
Consider a partner preparing for a client’s annual review. Once each system is connected to the firm’s AI tool, a single request can draw on Introhive to show who at the firm knows the client’s new CFO and whether any key relationships have gone quiet, while the billing system shows whether revenue has dropped and the practice management system shows which matters are open. Without those connections, assembling that picture would take an afternoon of emails.
Driving ROI: embedding vs. prompting
Most firms (63%) have no revenue-linked way to measure BD ROI, so AI spend gets judged on logins rather than outcomes. When IT and BD agree on two or three use cases, such as cross-sell, at-risk accounts, or pitch staffing, and connect the data those use cases need, you can track the revenue they influence.
In many firms, usage climbs for the first few weeks after go-live and then settles back to the same handful of enthusiasts. Some partners are already experimenting with tools like Claude on their own, but given that fee earners bill in six-minute increments, few will open a new portal and write a detailed prompt before every client call. An adoption plan that depends on partners learning prompt engineering is asking them to change how they work before they’ve seen what they get in return.
The firms that get past that point tend to see the difference in the work itself. Thomson Reuters Institute research found that partners who use AI daily across multiple types of work are nine times more likely to report a significant impact on efficiency and quality than those who use it minimally, which suggests the value builds as AI becomes part of how they work every day rather than something they reach for occasionally.
Deloitte Canada described this approach at our London customer summit, where its team explained that the pre-meeting digest, generated from meetings already on a partner’s calendar, has become the channel its partners open most. Embedding AI workflows into moments partners already have scheduled, rather than waiting for them to type a query, is what gets the insight to them while it can still shape the conversation.
Choosing how context reaches your AI tools
Connecting systems is not a new problem for IT, but composable AI legal tech changes what those connections are for. The traditional answer is a custom integration for each pairing, such as CRM to billing or a relationship intelligence API to a reporting tool, each with its own scoping, build, and maintenance. Those projects take months, and the pipelines they produce tend to break when a vendor changes its API or ships an update. Repeat that for every AI tool the firm adopts, and you’ve rebuilt AI tool sprawl at the integration layer.
The Model Context Protocol takes a different approach. MCP is an open standard that gives AI tools one consistent way to discover and use data from other systems. Instead of building a custom relationship intelligence API integration for each AI tool, you expose a system once through an MCP server, and each compatible tool your firm approves can be pointed at it through configuration rather than a new build. When your firm changes models or adds an agent, the server and the governance around it stay in place.
Many firms haven’t made that choice yet. When Introhive polled professional services leaders at a recent Solutions Spotlight session, the largest group (37%) said they hadn’t decided where their firm’s business context will live. Only 11% had chosen to connect it to AI tools through MCP or a similar protocol. Firms that decide early will spend less time rebuilding integrations later.
Connecting a system, however, is only half the decision. A connection that hands an AI tool raw rows from a database leaves the model to work out what they mean, and relationship data is especially hard to interpret that way, given that a contact record alone can’t say who at the firm knows a client best or whether that relationship is cooling. Where Introhive’s relationship intelligence API returns records for your developers to build with, the MCP Server returns answers rather than rows, drawing on relationship strength, recency, and engagement trends, so the AI tool receives the context behind the data rather than the data alone.
For firms already building a legal tech data lake, Introhive’s MCP Server works alongside that investment rather than in place of it. Introhive’s relationship intelligence API and data extracts can feed the lake for reporting and analytics, while the MCP Server gives compatible AI tools governed answers about the firm’s relationships. Because it’s model agnostic, your Model Context Protocol enterprise architecture isn’t tied to any one AI vendor.
MCP Server vs. API: how context reaches your AI tools
An API returns records that developers use to build a custom integration for each AI tool, while an MCP server connects a system once so any compatible tool the firm approves can be pointed at it through configuration. For professional services firms running more than one AI tool, that difference decides how much integration work each new tool adds.
| API / CUSTOM INTEGRATION | MCP SERVER | |
| What it is | A connection built for each pairing of systems (e.g., CRM to billing) | An open standard that gives AI tools one consistent way to discover and use data from other systems |
| Connecting a new AI tool | A new project with its own scoping, build, and maintenance | Connect the system once through an MCP server; each compatible, approved tool is pointed at it through configuration, not a new build |
| Effort over time | Projects take months, and pipelines tend to break when a vendor changes its API or ships an update | When the firm changes models or adds an agent, the server and its governance stay in place |
| What it returns | Records for developers to build with | Answers rather than rows, drawing on relationship strength, recency, and engagement trends |
| Best suited for | Feeding a data lake for reporting and analytics | Giving compatible AI tools governed answers about the firm’s relationships |
Uncompromising security and AI data governance
None of this matters if it can’t pass your security review. As Chad Ergun, CIO at Womble Bond Dickinson (US), notes in the Fall 2026 issue of ILTA’s Peer to Peer, the information firms connect to AI ranges from public contact details to sensitive material in internal records. That data needs to be classified and screened first, with its permissions and jurisdictional rules respected. Otherwise, more context creates risk rather than confidence.
Partners are open to connecting data to AI on one condition: the firm stays in control. In Introhive’s Partner Survey, 83% said yes or maybe to sharing email content with AI tools when the firm controls data governance. Client billing data remained the most sensitive category, especially in legal and accounting.
MCP is an access layer, not a new place your data lives. Your relationship data stays in Introhive, and the AI tool receives only the context needed to answer each question. Customer data is never used to train third-party AI models. Each user signs in through OAuth with their own Introhive account, and access generally mirrors the permissions they already have in Introhive. The connection is read-only, so that an AI tool can retrieve relationship intelligence but can’t create, change, or delete records. Only your Introhive admin can set up a connection, and only licensed users can use it. All of it runs on a platform backed by Introhive’s SOC 2 Type II attestation and ISO 27001 certification, and compliance with GDPR and PIPEDA.
The answer to AI tool sprawl isn’t another tool. It’s giving the tools you already have the same trusted relationship context and embedding AI workflows where your partners already work. A Model Context Protocol enterprise architecture is how firms are making both possible.
Want to see how Introhive connects your firm’s relationship intelligence to the AI tools you already use? Book a demo with our team.
BOOK A DEMO
