When we compared the top IP management systems, we judged them on one capability above all others: open connectivity. Intelligent automation was happening outside traditional systems, so the most important system was the one that let you connect to it.
That advice still holds, but it answers last decade’s question: which database will let us bolt AI on from the outside? The question for a modern IP firm is different. Is the system wired from the inside out so AI can do the work, all of the back-office work and not just docketing, under controls the firm sets and a record the firm can audit?
OttoFlow™: Badlands Edition was built to answer yes. It comes from the team that built Foundation IP, the application with the largest law-firm customer base in the field. In OttoFlow, automation is not wired to the system. Automation is the system. And because AI is only as useful as the authority you can safely give it, OttoFlow is just as deliberate about what the AI may not do.
The Ceiling on Bolted-On AI
Every firm can now point a capable model at its work. The trouble is what that model can reach. In a traditional IPMS, AI sits outside the record. It can summarize a document someone hands it, or answer questions through a vendor’s chat feature. It cannot safely write a docket entry, open a workflow, report a citation, or instruct a foreign agent, because the system was never designed to let software act on the record with a person’s authority and a person’s accountability.
So the people still do the work. Dates are re-keyed from correspondence. Citations are tracked in spreadsheets. Client reports are assembled on request. A few people carry the firm’s conventions in their heads, and quarter-end runs on long nights. The AI helps around the edges, but the back office does not change.
Meanwhile, the corporate IP departments behind the largest portfolios have raised the bar. They expect fewer human touches per office action, docket data their own AI tools can reach, and a security questionnaire answered with mechanisms rather than policies. A system that stores the work cannot meet those expectations. A system that does the work can.
What “Inside Out” Means
A system wired for AI from the inside out has three properties. OttoFlow was designed around all three.
- Every event is an input. OttoFlow is event-driven from the ground up. Incoming PTO correspondence, examiner actions, foreign agent reports, and inbound email are matched to the right matter by application number and document code the moment they arrive. Nothing waits for a person to notice it.
- Every record is in one schema. Bibliographic data, file wrappers, continuity, citations, and examiner analytics are pulled from the patent offices into a single schema. USPTO data arrives through Otto HUB™ and Patent Center, alongside Global Dossier, EPO OPS, WIPO, JPO, KIPO, and CNIPA. AI reasons over the whole matter, not a document someone pasted in.
- Every action has a governed path. A scoped, idempotent, Swagger-documented Agent API covers the whole record. CloudEvents webhooks and DMS push carry results outward, and a built-in MCP controller lets Claude, ChatGPT, and Microsoft Copilot work inside OttoFlow. AI acts through the same validated operations people use, never around them.
The first two properties make AI capable. The third makes it trustworthy. Together they let a firm move AI from answering questions to doing work.
AI Across the Whole Back Office
Docketing is where OttoFlow proved the model. Its rule library holds 1,853 docketing procedures across 63 jurisdictional rule sets, with 205 jurisdictions seeded, and the engine has run more than five million fully automated docketing transactions. The same architecture reaches every other part of the back office:
- Docketing. Documents are matched, run through a versioned rule library with three-tier client overrides, and turned into due dates computed from country law and holiday calendars. Rules and AI check the result, and a person confirms by exception.
- Correspondence. Inbound email lands in the correspondence queue already matched by application number. The document store OCRs what arrives.
- Prosecution. AI-enabled notice-of-allowance processing, shell creation, and error surveillance remove chores, not judgments. A full merge-document library produces the firm’s standard documents from the record.
- IDS and prior art. IDS Manager reads PTO-892, 1449, and SB/08 forms, keeps a master reference list, propagates citations across the family, maintains cross-cite pools, and generates the SB/08. The same engine computes IDS deadlines.
- Drafting. Otto IP™, the attorney-built AI for applications and responses, runs on the same platform and draws on the same record.
- Foreign filing. The Foreign Filing Hub turns a filing decision into instructed, tracked work with more than 300 agents through Honu™.
- Client reporting. Reports are generated on a schedule from the record, in the client’s own format, instead of being prepared by staff on request.
- Corporate programs. Invention disclosures, inventor remuneration (including the German statutory formula), and LEDES and UTBMS billing are built in, on the same record the AI can reach.
- Research and triage. Matter Search is built for 20,000-matter portfolios and takes natural-language queries. Attorneys can also ask OttoFlow questions from Claude, ChatGPT, or Copilot.
- Onboarding and migration. Import rules read a transfer package and suggest field mappings with AI, with audit history and rollback.
Because every one of these functions sits on the same schema and the same governed action paths, a firm does not have to buy, integrate, and secure a separate AI tool for each one. It can apply AI wherever the work is, and that includes AI the firm already uses.
Every Application of AI, Under Fine-Grained Control
Opening the whole back office to AI is only responsible if the firm controls exactly what the AI can touch and what it may do on its own. OttoFlow treats that control as part of the architecture, not a policy memo. It is layered:
- Which AI. Firms connect their own AI provider keys rather than being tied to a vendor’s model. Claude, ChatGPT, and Copilot connect through the MCP controller, and Otto IP handles drafting. Each of OttoFlow’s eleven AI features is enabled per tenant, so a firm turns on exactly the capabilities it is ready for.
- What it can reach. The Agent API is scoped, and 18 permissions across eight domains define what any user, and any AI acting for that user, can see and change. An AI never holds more authority than the person it works for.
- What it can do on its own. Every AI write is validated against the matter, the obligation, the rule, and the user’s privileges before it lands. Routine, rule-backed work completes automatically. Anything the AI sourced is labeled and routed to a dedicated verification queue for a person to confirm.
- How much it can do. AI features are metered and governed by rate and spend limits, so autonomy never turns into a runaway bill or a runaway batch.
- Who answers for it. AI-entered work is audited as AI-entered under the user’s name, in an append-only audit stream. Bulk actions carry undo, and AI-assisted imports carry rollback.
The result is a dial, not a switch. A firm can start with AI that reads and suggests, move to AI that acts on routine work while people verify by exception, and widen that scope one feature, one role, and one client at a time, without changing systems.
Delegation with Supervision
This is how good firms already work with people. A partner delegates to an associate, an associate to a paralegal, and quality comes from clear scope, a defined review step, and a record of who did what. OttoFlow applies that same discipline to AI.
The AI is given a scope: the features enabled, the permissions held, and the actions it may complete without review. Its work lands where the team already works, in the role-scoped Firm Docket, personal queues, and digests, with AI-sourced items set apart in the verification queue. People confirm by exception rather than redoing the work, and judgment calls stay with the attorney. When someone asks what happened, the audit trail shows what the AI did, under whose authority, and who checked it.
That is the difference between AI you experiment with and AI you can delegate to. Supervision is built into the workflow, so quality control does not depend on someone remembering to look.
Quality Control as a Record, Not a Promise
Thoroughness in prosecution is mostly a question of what the system makes you prove. In an hours-based model, a docketed date carries a user name and a timestamp. In OttoFlow, every date carries its source document, the rule that produced it, the formula, the base date, and the person who verified it. When a client asks where a deadline came from, whether a person or an AI produced it, the answer is already on the record.
Consistency is built the same way:
- Deadlines come from a versioned, tested rule library, so the same document produces the same date every time.
- Workflow templates are versioned and carry stages, tasks, checklists, and budgets, so every step is defined and visible.
- Six alert types reach the right people by role, by email, in-app, or in a digest.
- Error surveillance watches for problems rather than waiting for someone to notice them.
- The matter record shows what remains uncited, so IDS gaps are visible before they become problems.
And when the corporate client’s security questionnaire arrives, OttoFlow answers it with mechanisms: tenant isolation enforced by PostgreSQL row-level security, TOTP multi-factor authentication, SAML 2.0 and OIDC single sign-on, encryption in transit and at rest, and hosting on AWS with region pinning available.
The Economics Follow from the Design
Legacy vendors grew by selling docketing hours, often from offshore service teams. Their business depends on work staying manual. OttoFlow is licensed as software only, so every step that runs itself is the product doing its job, not a line on an invoice. As a firm widens the scope it delegates to AI, its costs go down, not up.
How OttoFlow Stacks Up
By the standard we set in our earlier comparison, open connectivity first, OttoFlow would top the list. Judged by how much of the back office AI can safely run, the gap is wider.
| What we looked at | The eight systems we compared | OttoFlow: Badlands Edition |
|---|---|---|
| Connectivity | Ranged from strong web-services APIs to none at all; three of the eight had no API | Scoped, idempotent Agent API across the whole record, CloudEvents webhooks, DMS push, and a built-in MCP controller |
| Where AI sits | Outside the system, or as a chat feature inside the vendor’s product | Inside the system, acting on the record through the same validated operations people use |
| Scope of automation | Docketing, at best, and usually keyed by people | Docketing, correspondence, prosecution chores, IDS, drafting, foreign filing, reporting, and onboarding |
| Control over AI | Bounded by the vendor’s own chat feature, where one exists | Per-tenant feature enablement, scoped permissions, the firm’s own model keys, and rate and spend limits |
| Supervision of AI work | Left to the firm’s own review habits | AI-sourced work labeled and routed to a verification queue; people confirm by exception |
| Accountability | A user name and a timestamp | Source document, rule, formula, base date, and verifier on every date; AI entries audited under the user’s name |
| Licensing model | Software, often with docketing hours sold on top | Software only; every step that runs itself is the product working |
The first row is where our earlier comparison stopped. Every row after it is work a traditional system leaves to the firm, and risk a traditional system leaves unmanaged.
See Your Own Back Office Run Itself
The fastest way to evaluate OttoFlow is not a feature checklist. It is your own mail. Pick one corporate client and send us a month of their correspondence. We process it on a pilot tenant set up with that client’s fields and conventions. You then walk through what ran on its own, what the AI routed for verification, and how each date traces to its document, rule, and formula. Then you decide on a parallel pilot and how much to delegate.
Nothing has to move at once. Host-system synchronization keeps OttoFlow current with a legacy platform during a phased migration, and firms that keep their docket elsewhere can start with OttoFlow IDS Manager and bring citation management and SB/08 generation in first.
Schedule a demo, or let’s talk about how much of your back office you’re ready to delegate.