In the last post in this series, we made the case that AI’s real value for solo attorneys isn’t writing briefs - it’s handling the 48% of your day that goes to administration. Time entry review, invoice drafting, email triage, deadline monitoring. The operational work that eats your hours and leaks your revenue.
But we left out the hard part: how does any AI tool actually do those things for your practice?
The answer is deceptively simple and critically important: the AI has to be able to see your data. And right now, for most solo attorneys, it can’t.
The Copy-Paste Problem
Here’s what “using AI” looks like for most attorneys today.
You open ChatGPT. You type: “Draft an invoice narrative for the Thompson estate matter.” ChatGPT asks what work you did. You switch to your calendar, find the meetings. Switch to your email, find the correspondence. Switch to your time tracking tool, find the entries. Copy it all. Paste it into ChatGPT. Wait for the output. Copy the output. Paste it into your billing software.
You just spent ten minutes doing data entry so that AI could save you five minutes of writing. The math doesn’t work.
This is where 57% of solo attorneys who’ve tried AI are stuck right now - using generic tools like ChatGPT that have no connection to their practice data. The AI is smart, but it’s blind. It doesn’t know your clients, your matters, your time entries, or your calendar. Every interaction starts from zero.
The attorneys getting real value from AI aren’t smarter about prompting. They’re using AI that’s connected - tools that can query their practice management system directly, see what’s there, and act on it without the copy-paste relay.
What MCP Actually Is (and Why You Should Care)
MCP stands for Model Context Protocol. If that sounds technical, here’s the plain version: it’s a universal connection standard that lets AI tools talk to your software.
Think of it like USB. Before USB, every device needed its own proprietary cable. Your printer had one plug, your keyboard had another, your external drive had a third. USB created a single standard - one port, any device. You stopped thinking about cables and started thinking about what you wanted to do.
MCP does the same thing for AI. Instead of every AI tool building a custom integration with every software platform - Copilot needs a special connector for your billing software, Claude needs a different one, ChatGPT needs yet another - MCP creates one standard connection. If your practice management software supports MCP, any AI tool that speaks MCP can connect to it. Your matters, your time entries, your clients, your calendar - accessible to whatever AI you choose to use.
This isn’t theoretical. It’s working today. When we showed Microsoft Copilot finding unbilled time by scanning a Writly user’s Outlook calendar, OneDrive files, and email activity, that connection ran through MCP. Copilot asked Writly for the user’s matters and existing time entries. Writly responded through the MCP server. Copilot compared that to the user’s Microsoft 365 activity and found the gaps. No copy-paste. No manual data entry. No switching between tabs.
Here’s what this looks like in practice:
See the full breakdown: How Copilot Finds the Billable Hours You Missed
Why This Matters More for Solos Than Anyone Else
Large firms have IT departments that build custom integrations. If BigLaw wants their AI tool connected to their document management system, they hire developers to make it happen. The integration is proprietary, expensive, and locked to that firm.
Solo attorneys don’t have that option. You need standards-based connections that work out of the box - the same way you need a printer that works when you plug in the USB cable, not one that requires a custom driver written by a consultant.
MCP is that standard. And it matters for solos specifically because of three dynamics:
You can’t afford to be locked in. The AI landscape is moving fast. The best tool this year might not be the best tool next year. If your practice management software only connects to one AI vendor’s product, you’re stuck with whatever that vendor decides to build. If it supports MCP, you can switch AI tools without switching practice management platforms. You can use Copilot for time entry review today and Claude for invoice drafting tomorrow - same data, different tools, no migration.
Your data is your competitive advantage. A solo attorney’s practice data - the patterns in their billing, the relationships between matters and clients, the workflow habits that drive revenue - is uniquely valuable when an AI can analyze it. But that value only exists if the AI can actually access it. An open connection turns your practice data into a working asset. A walled garden keeps it locked in a silo where only one vendor’s AI can touch it.
The admin burden falls entirely on you. At a firm, a paralegal handles time entry cleanup and a billing department assembles invoices. In your practice, that’s all you - and it’s costing you 25-40% of your potential revenue. AI connected to your practice data can do that work. AI that requires you to manually feed it information is just adding another step to the process.
What “Connected AI” Actually Does in Practice
Let’s make this concrete with three scenarios that work today - not future-state product roadmap items, but things an attorney with a connected setup can do right now:
End-of-day time audit. You ask your AI: “What did I work on today that I haven’t billed?” The AI checks your calendar (two client meetings), your email (four substantive client exchanges), and your document activity (one contract review, one letter drafted). It compares that to the time entries already logged in your practice management system. It finds 1.5 hours of work you forgot to record. It drafts the time entries - matter, description, duration - pre-filled and ready for your review. You approve them in 30 seconds.
Without a connection: you’d need to manually check your calendar, scan your sent mail, try to remember what documents you worked on, cross-reference with your existing time entries, and type it all in. That’s 15 minutes of admin work, and you probably still miss something.
Invoice preparation. You tell your AI: “Generate the invoice narrative for the Martinez matter.” The AI pulls all unbilled time entries for that matter, organizes them chronologically, drafts a professional narrative summarizing the work, flags an entry that looks unusually short for the task described, and presents the draft for your review. You edit one line and approve.
Without a connection: you export time entries, open a document, write the narrative from scratch referencing the raw entries, format the invoice, and copy it back into your billing system.
Monthly practice health check. You ask: “How did my billing look last month?” The AI analyzes your time entries, invoices, and payments across all matters. It tells you your utilization was up 12% month-over-month, your average time-to-payment dropped by three days, two matters haven’t been billed in 45 days, and your most profitable practice area last month was estate planning. This is the kind of analysis most solos never do because pulling and crunching the data takes hours. With a connected AI, it takes seconds.
The Walled Garden Problem
Not every practice management platform is building toward openness. Some are building walled gardens - proprietary AI features that only work within their ecosystem. We compare open vs. closed approaches to practice management in our buyer’s guide.
The pitch sounds good: “Our AI does everything you need, built right in.” But here’s what that actually means:
You get the AI features that vendor decides to build, on the timeline they decide to build them. If their AI can draft invoice narratives but not review time entries, you wait. If their AI works with their email integration but not your actual email provider, you’re out of luck. If a better AI model comes along next year, you can’t use it - you’re locked to whatever model your vendor chose to embed.
The walled garden also means your data has one exit: leaving the platform entirely. Your practice data, your billing history, your client relationships - they feed the vendor’s AI, but you can’t take that data to an independent AI tool for a second opinion or a different kind of analysis.
An open platform with MCP support inverts that dynamic. Your data stays in your practice management system. Any AI that supports MCP can access it - with your permission, under your control. You’re not choosing an AI vendor when you choose your software. You’re choosing a platform that lets you use whatever AI works best for each task.
What to Look For
If you’re evaluating practice management software - or re-evaluating what you already have - here’s what matters on the AI connection front:
Does it support MCP or an equivalent open standard? If the answer is “we have our own AI built in,” ask the follow-up: can I connect other AI tools to my data? If the answer is no, that’s a walled garden.
What data is exposed through the connection? Matters, contacts, time entries, billing, and calendar are the minimum. The more of your practice data the AI can see, the more useful it becomes.
Who controls access? You should be able to decide which AI tools connect, what data they can access, and when. An MCP server that exposes everything to everything isn’t secure. One that lets you grant specific permissions to specific tools is.
Can you use it today, not “coming soon”? A lot of platforms are talking about AI integration. Fewer have actually shipped it. Ask for a demo, not a roadmap.
The Bottom Line
AI is only as useful as the data it can see. For solo attorneys, that means the gap between “I use ChatGPT sometimes” and “AI runs my back office” comes down to one thing: whether your AI can connect to your practice data without you playing middleman.
MCP is the open standard that makes that connection possible. It’s not the only path - but it’s the one that doesn’t lock you into a single vendor’s AI roadmap.
The attorneys who set up this connection now won’t just save time. They’ll capture more revenue, send cleaner invoices, and finally do the practice analytics they’ve been meaning to get around to for years. And they’ll be able to switch AI tools as the market evolves, without starting over.
Your practice data is the most valuable business asset you have. Make sure your AI can actually use it.
This is the second post in “The AI-Powered Solo Practice” series. Previously: Stop Worrying About AI Writing Your Briefs. Start Using It to Run Your Office. Next: “The Solo’s AI Toolkit: What to Use, What to Skip, and What to Wait For.”
CTA: Writly’s MCP server connects your practice data to Copilot, Claude, ChatGPT, and any AI tool that supports the standard. Your data goes wherever you want to take it. $27/month for Founding Firms. Start your free trial →
Starting a solo practice?
Writly was built for this moment — practice management for solo attorneys, $27/month for Founding Firms, send your first invoice in an hour. Everything you need from day one, nothing you don't.
Start your free trial →