[ FILEVINE INTEGRATION ]
Filevine is your system of record. We build the custom API integrations, medical-record review, demand-letter drafting, and privilege-safe Claude workflows it doesn't ship — on top of the Filevine API, with ABA Opinion 512 audit logging and attorney verification. You own the result.
30 min with a co-founder. No sales pitch.
4–6 wk
build phase
ABA 512
audit logging built in
Free
discovery call first
[ WHAT WE BUILD ]
Filevine runs your projects, documents, and deadlines. We build the custom layer it doesn't ship — integrations, custom reporting, medical-record review, and privilege-safe AI — against the Filevine API, and you own the result.
Custom apps, data syncs, and webhook automations built directly on the Filevine REST API — projects, sections, documents, contacts, tasks, deadlines. Connect Filevine to your billing, e-sign, and lead sources. Data stays on your infrastructure.
AI that extracts and summarizes medical records into the right Filevine project section — chronologies, treatment summaries, and paid-versus-incurred analysis. Every summary flagged for attorney review before use.
Draft demand letters generated from case data and your templates. Claude produces the first draft from the Filevine project; attorneys review and approve before anything goes client-facing.
Automated intake and lead routing that opens a Filevine project with fields pre-populated, plus deadline and statute-of-limitations triage so nothing slips. Integrates with your existing intake forms and lead sources.
Agents that run multi-step firm workflows on Filevine data: document review, section updates, deadline triage. Every action logged append-only and gated behind attorney approval.
The reports the native Report Builder can't produce: cross-collection views, phase aging, demand aging, SOL risk, staff throughput, settlement realization. Built on DataBridge/Snowflake into Power BI or Looker Studio for standing dashboards, or on the API for targeted reporting. Read the reporting guide.
Duplicate rows, wrong totals, and exports that time out usually trace back to a migration that pasted the old database into Filevine without normalizing it. We run a field-mapping audit, consolidate fragmented fields into clean collection sections, and restore reporting integrity.
Ask "which cases have had no note activity in 30 days" and get an answer, no report built. Our open-source Filevine MCP server (@oktopeak/filevine-mcp on npm) connects Claude to your projects, contacts, deadlines, and custom collections, with schema discovery for your firm's sections.
Custom deployments where the model runs on your own infrastructure and nothing leaves the firm. Works with any LLM, not just Claude. For firms whose confidentiality requirements rule out third-party AI processors. Read the Privilege Stack guide.
[ THE PROBLEM ]
Filevine is the system of record for many PI and mass-tort firms. But the integrations and AI workflows on top of it are yours to build — and most attempts fail in three predictable ways.
01
02
03
[ PROCESS ]
We de-risk a new platform with an early integration spike, so you see real Filevine data flowing before the full build.
A 30-minute call with a co-founder. We map how your firm runs on Filevine, find the 3-5 highest-impact automations, and give you a draft scope and rough quote. No charge, no obligation.
Before the full build, we run an early spike against the Filevine API so you see real data flowing. It de-risks a new platform and proves the workflow before you commit to the whole scope.
4-6 weeks to build, plus 8 weeks of embedded support during rollout — because 95% of AI pilots fail without structured implementation support. Fixed price from the discovery scope. You own the source.
[ WHAT WE'VE SHIPPED ]
We haven't shipped a public Filevine project yet, and we're upfront about that. Filevine is a new instance of the same engineering behind our open-source connectors and regulated builds — you can audit that work today.
Open Source
Clio & MyCase MCP Connectors
Open-source Claude connectors for the two practice-management systems most US firms run on. OAuth, append-only ABA 512 audit logging, published to npm. The same engineering pattern we bring to Filevine.
8 weeks
DEA Compliance Platform
A regulated build with an immutable, append-only audit trail. 100% automated Form 41 processing, biometric authentication, encrypted video witnessing, 65+ API endpoints. Passed DEA inspection.
$220K/yr
Legal Document Search Platform
Elasticsearch on 1M+ legal documents. Sub-100ms queries with legal terminology handling and fuzzy matching. Recovered $220K+/year in attorney search time.
[ COMPLIANCE ]
Every Filevine workflow we build satisfies the ABA's AI ethics requirements. Not as an afterthought.
AI-generated demand letters, medical summaries, and pleadings require attorney review before any client-facing use. No automated output bypasses human judgment.
Every AI interaction logged: prompt, response, source documents, model version, timestamps, and reviewer identity. Append-only. Court-defensible.
Enterprise-tier AI only. Training disabled, Zero Data Retention. Reached through the Filevine API, not the chat window. Not used for training, not stored after the response.
[ FAQ ]
Yes. Filevine exposes a REST API plus webhooks that let you read and write projects, sections, documents, contacts, tasks, and deadlines. That's the foundation for custom integrations, data syncs, and automations on top of Filevine.
We build directly against the Filevine API rather than relying on off-the-shelf connectors, so the integration matches your firm's actual workflow.
Yes. We build privilege-safe AI workflows that read from and write to Filevine through its API: medical-record summaries dropped into the right project section, draft demand letters generated from case data, intake routing, and document review. The AI reads case data through the API rather than a chat window, runs on enterprise-tier AI with training disabled and Zero Data Retention so content is not used for training or stored after the response, and every AI action is logged append-only for ABA Opinion 512.
No. A 2026 federal court ruling found that information entered into public AI platforms is not protected by attorney-client privilege. Pasting medical records, demand drafts, or client communications into consumer ChatGPT or Claude can waive privilege.
We build on enterprise-tier AI with training disabled and Zero Data Retention, reach Filevine through its API rather than the chat window, and implement audit trails that satisfy ABA Opinion 512 requirements. Content is still processed by Claude, but under those terms it is not used for training or stored after the response. See our breakdown of the privilege ruling.
Common Filevine automations: demand-letter drafting from case data, medical-record extraction and summarization into project sections, intake and lead routing, deadline and SOL triage, pleading drafts from templates, and data syncs between Filevine and your other systems (e-sign, billing, lead sources, document management).
Every AI-generated output is flagged for attorney review before client-facing use.
Simple integrations and automations are scoped on a short call. Custom builds — document automation, medical-record review AI, intake routing, custom agents, full compliance architecture — run from $12K depending on the number of workflows and the depth of integration required. Deploying the connector for a whole team rather than one attorney is its own scoped engagement: one setup, every attorney.
Discovery is free: a call, a draft scope, and a rough quote. We provide a fixed-price proposal within five business days. Book a 30-minute call.
4-6 weeks for the build phase, plus 8 weeks of embedded support. The support phase matters because 95% of AI pilots fail without structured implementation support. We stay embedded during rollout to ensure your team actually adopts the workflows.
No. Filevine stays your system of record. We build integrations, AI workflows, and automations on top of it through the Filevine API. No migration, no data export, no disruption to your case teams. The work layers onto the platform you already run.
Yes, and it's public. We built and published the open-source Filevine MCP server (@oktopeak/filevine-mcp on npm), alongside our connectors for Clio and MyCase — so you can read our Filevine integration code before you call us.
That build covers the parts of the Filevine API that trip up most integrators: the personal-access-token auth flow, the three-header request model, per-firm custom collection sections with schema discovery, and the rate limits around reporting endpoints. We wrote up the full reporting landscape in our Filevine custom reporting guide.
Every workflow we build includes a verification layer. AI-generated demand letters, medical summaries, and pleadings are flagged for attorney review before any client-facing use. Audit trails log every AI interaction, every source document referenced, and every human approval. This satisfies ABA Opinion 512 competence and verification requirements.
Yes. For firms whose confidentiality requirements rule out third-party AI processors, we deploy models that run on your own infrastructure so nothing leaves the firm. The same privilege-safe Filevine integration pattern works with on-prem or self-hosted models, not just hosted Claude. Read the Privilege Stack guide.
[ GET STARTED ]
30 minutes with a co-founder. We'll map your highest-impact Filevine automations and tell you honestly what's worth building. Discovery is free, including a draft scope and a rough number.
