Best MyCase + Claude integration solutions in 2026: an honest comparison

Law firms ask us this every week: "How do I integrate Claude with MyCase?" There are five practical paths in 2026, and they are not interchangeable. Pick the wrong one and you will either spend nine months in compliance limbo or pay five times more than necessary.

This is an honest comparison of all five, written by the team that built the first open-source MyCase MCP connector. Yes, we have a horse in this race. We have also worked alongside firms using every other option on this list, and we will tell you where ours is not the right fit.

How we evaluated each option

Six criteria, weighted toward what matters most for real firms with active matters and bar oversight:

• Open source / readable code. Can your IT lead read what the integration actually does with your client data? This matters more than most firms realize until a state bar starts asking questions.
• ABA Opinion 512 audit logging. Does the integration log every AI tool call in a form your bar would accept on audit? Logging is the difference between "we used AI" and "we used AI responsibly."
• Setup complexity. Hours from "we want this" to "this is running on a partner's machine."
• Production readiness. Will it survive a real firm's daily case volume, or will it break the first time someone uploads a 200-page deposition?
• Cost (TCO). 12-month all-in cost for a 10-attorney firm, including the Claude subscription cost where applicable.
• Scope of capabilities. General-purpose AI workflow (research, drafting, summaries, automation) vs. single-purpose (intake, receptionist, billing).

Each option below gets a frank assessment on all six. Pricing references are accurate as of May 2026 and will move.

Option 1: Oktopeak MyCase MCP connector (open-source)

What it is: An open-source Model Context Protocol server that lets Claude Desktop or Claude Code read and write to your MyCase case management system directly. 18 tools across cases, contacts, documents, tasks, calendar, calls, staff, time entries, and billing. Append-only audit logging that maps to ABA Opinion 512 requirements out of the box. MIT licensed. Available on GitHub, npm, and the official Anthropic MCP Registry.

Best for: Firms that have at least one technical lead who can configure an MCP client. Firms that want full Claude capabilities (research, drafting, summarization, multi-step workflows) rather than a single narrow use case. Firms under any kind of compliance scrutiny who need a real audit log, not a checkbox.

What we did well:

• You can read the source code before you trust it. Every tool call, every auth flow, every encryption choice is in our public GitHub repo.
• Every Claude interaction with your MyCase data is logged append-only at ~/.oktopeak-mycase/audit.log with timestamp, tool name, arguments, and result. This is the artifact your bar would request.
• OAuth tokens are encrypted at rest with AES-256-GCM. Stored locally in your OS keychain. Never sent to our servers.
• Free under MIT. Will stay free.

Where it does not win:

• Requires a developer or IT-comfortable partner for initial setup. We say "five minutes once your MyCase API credentials are in hand," but that assumes you can edit a JSON config file in Claude Desktop. If that sentence intimidates you, you want Option 4 (Zapier) or a paid setup.
• It is a connector, not a finished workflow. You still need to design what you want Claude to do with the data.
• Requires the MyCase Advanced tier (currently $109 per user per month) because MyCase gates Open API access to Advanced. This is a MyCase pricing decision, not ours.

Cost: Free (open-source) plus Claude Pro at roughly $20 per user per month or Claude API pay-per-token. For a 10-attorney firm: $0 for the connector, $200 per month for Claude Pro across all attorneys. If you want us to deploy it for you with a custom workflow, book a call to scope it.

Option 2: Clerx AI

What it is: A MyCase native marketplace integration focused on AI-powered client intake. Automates inbound lead qualification, screens potential clients, and routes qualified matters into MyCase with pre-populated case records.

Best for: Firms whose biggest pain point is the intake pipeline. Personal injury, immigration, family law, and similar practice areas with high inbound volume and qualification overhead. Teams that want a configure-it-once-and-forget product, not a general AI assistant.

What it does well:

• Tight native integration with MyCase. Listed in the MyCase marketplace at mycase.com/integrations/clerx-ai/. Setup goes through MyCase's normal integration flow rather than requiring developer work.
• Purpose-built for intake. Forms, qualification logic, and case routing are all pre-designed for the typical law firm intake funnel.
• Vendor-managed. No infrastructure on your side.

Where it does not win:

• Intake only. Cannot help with case research, document drafting, deposition summarization, billing review, or any general-purpose Claude workflow.
• Closed-source. You cannot read the code that processes your prospective clients' information.
• If your AI needs grow beyond intake, you will end up running Clerx AI alongside one of the other options on this list anyway.

Cost: Pricing is not publicly listed on the MyCase marketplace page as of May 2026. Contact Clerx AI directly for a quote.

Option 3: Smith.ai

What it is: 24/7 AI-powered receptionist and chat service that handles inbound calls and web chats for law firms, routes qualified leads into MyCase, and books consultations on your calendar. Combines AI agents with human escalation for nuanced calls.

Best for: Solo practitioners and small firms that lose leads to missed after-hours calls and unanswered web chats. Firms whose entire problem is "we cannot answer the phone fast enough."

What it does well:

• Covers the channel most law firm AI integrations ignore: the phone.
• Native MyCase integration via the marketplace.
• Human-in-the-loop for calls that require nuance. Pure AI agents alone routinely mishandle legal intake; Smith's hybrid model handles that well.
• Time-to-value measured in days.

Where it does not win:

• Not a general-purpose Claude integration. Will not help your associates with case research or your paralegals with document drafting.
• Pricing scales with volume. High-volume firms see meaningful monthly costs.
• You are renting capability, not owning it. If you cancel, you are back to missing calls.

Cost: Smith.ai pricing is published on their website and starts in the low hundreds per month for the receptionist plans, scaling with call and chat volume. A 10-attorney firm with active inbound typically lands in the $700 to $2,000 per month range, all-in.

Option 4: Zapier MyCase + Anthropic Claude integration

What it is: A no-code automation platform connecting MyCase events to Claude API calls via prebuilt zaps. Available at zapier.com/apps/mycase/integrations/anthropic-claude. Trigger examples: "new lead in MyCase → Claude drafts intake summary," "case updated → Claude generates status email."

Best for: Firms with no developer on staff who want a specific narrow workflow up quickly. Teams that already use Zapier for other internal automations and want one more workflow added. Pilot use cases where you want to validate the idea before investing in a real build.

What it does well:

• No code. A legal ops person who has built Zapier flows for other tools can build a basic Claude + MyCase flow in an afternoon.
• Visible logging at the Zapier level. You can see which zap ran, when, and what data passed through.
• Cheap to start. Zapier Starter from $19 per month plus Claude API costs.

Where it does not win:

• No tool-call-level audit logging. Zapier logs zap runs, but the actual Claude prompt and response are not captured in a form that satisfies ABA Opinion 512 audit-log expectations. You would need to add custom logging in a second zap.
• Limited to what Zapier's MyCase connector exposes. Many MyCase API endpoints are not available through Zapier.
• Cost scales fast with volume. A firm running thousands of automation events per month exits Zapier's affordable tiers quickly.
• Brittle in production. Zaps fail silently if a field name changes upstream. Not designed for "this runs every day in a regulated workflow."

Cost: Zapier from $19 per month (Starter), scaling with task usage. Claude API costs separately, typically $20 to $80 per month for a low-volume firm. 12-month all-in for a 10-attorney firm running modest automation: $500 to $1,500.

Option 5: Custom build

What it is: Hire a developer or agency to build a tailored MyCase + Claude integration from scratch. Examples include specialized legal tech consultancies like Legal Pipeline (legalpipeline.dev), MoeCloud Group, and small dev shops working in the legal AI space. Includes our own paid Legal AI Integration service for firms graduating beyond the open-source connector.

Best for: Firms with a specific multi-workflow vision that off-the-shelf options cannot deliver. Firms with budget for a tailored build that they will own. Firms that have already validated the use case with one of the cheaper options on this list and want to scale it up properly.

What it does well:

• Exactly your workflow. No compromise on what the integration does or how it presents data.
• You can specify the compliance architecture from day one (audit logging, encryption, access control).
• You own the result. Source code, deployment scripts, walk-away documentation.
• Compounds. The first custom workflow makes the second cheaper.

Where it does not win:

• Cost. Production engagements are scoped per firm and are not cheap. Firms expecting a weekend's work at a weekend's price will be disappointed.
• Time to first value. Eight to twelve weeks is typical, even for compressed engagements. A firm with an urgent productivity gap should validate first with a cheaper option.
• Implementation risk. Bad agencies ship broken AI integrations that compound technical debt. Vet carefully: ask for their open-source contributions, audit log architecture, and three references in the legal vertical.
• You are responsible for ongoing maintenance unless you scope a retainer.

Cost: Scoped to the firm. Most agencies offer a paid discovery sprint to scope the work before quoting the build. Ask about their discovery process before signing anything larger.

Comparison table

Which one should you pick?

The honest decision tree, based on what we see in actual firm conversations:

• You have an IT lead or developer-comfortable partner, and you want general Claude capability: Start with the open-source MyCase MCP connector. Free, audit-ready, full Claude capability. If setup overwhelms your IT, add a Guided Setup.
• Your bottleneck is intake leakage: Clerx AI. It is purpose-built for the exact problem and will solve it faster than any general-purpose integration.
• Your bottleneck is missed calls and after-hours chats: Smith.ai. The phone channel is uniquely hard to automate well; their hybrid AI plus human model handles it.
• You have no technical staff and want one narrow workflow up quickly: Zapier. Validate the idea cheaply, then revisit when you outgrow it.
• You have a multi-workflow vision and a real budget: Custom build. Just vet carefully and demand to see prior audit-log architecture and references in the legal vertical.

What we actually recommend most firms do

Start free. Run the open-source MyCase MCP connector for two weeks with one of your partners actively using it on real (non-privileged) matters first. Learn what Claude is actually good at on your specific case data. The cost is zero plus a Claude Pro subscription.

If after two weeks you have a workflow you want to formalize, three paths open up:

• Your IT lead operationalizes it for the whole firm. Zero additional cost.
• You bring us in for a Guided Setup — we deploy it properly across the firm with scoped credentials, audit logging wired into your stack, and training. Scope and pricing are tailored to your firm.
• You bring us or another agency in for a full multi-workflow build. Bigger engagement, scoped to your firm.

The thing not to do is pay for a custom build before you have validated the use case on free tools. The most common pattern in expensive AI projects that fail is "we knew we wanted AI, we did not know what we wanted it to do." Two weeks with the open-source connector eliminates that risk.

One more thing about ABA Opinion 512

ABA Formal Opinion 512 (2024) requires attorneys using AI to supervise outputs, maintain client confidentiality, and document how the tool handles client information. Of the five options above, only the open-source MyCase MCP connector and a properly specified custom build ship with append-only audit logging that maps directly to those obligations.

This is not a small detail. The day a state bar or opposing counsel asks "show me your AI audit log for this matter," you will want a file. Not a memory.

Whatever option you pick, ask the vendor in writing: "What audit log do you produce when Claude touches my client data, and where is it stored?" If the answer is vague, treat it the same way you would treat a vague answer about malpractice insurance.

Disclosure and methodology

We are Oktopeak and we built the open-source MyCase MCP connector. We obviously prefer it. We also have not been paid by any other vendor on this list to position them favorably or unfavorably. The frank assessments come from working with firms using each of these options and from our own use of Claude across legal data.

Pricing references reflect publicly published prices and vendor quotes received in conversations with firms as of May 2026. Quotes will move; verify directly with each vendor before making a buying decision.

If you spot something inaccurate, email us at office@oktopeak.com and we will correct it.