SimProf: What a Flight Simulator Teaches About Patent Judgment

SimProf: What a Flight Simulator Teaches About Patent Judgment

In the last article, we explained why SLW built SimProf. This one goes inside a single session: what a summer clerk actually sees, the decisions they have to make, and the judgment the exercise is built to develop. 

A flight simulator does not exist to teach pilots how to fly in clear weather. It exists for the engine failure on takeoff, the crosswind landing, the instrument panel that starts lying to you. The whole value is in the situations you cannot safely practice on a real aircraft with real passengers aboard. 

SimProf is SLW’s in-house patent-prosecution training platform, and it runs on the same principle. For anyone meeting SimProf for the first time, SimProf is organized into several modules that mirror the real work of prosecution:  

• Preparation: drafting claims from invention disclosures and prior art; Prosecution, responding to office actions with amendments and argument 
• Simulation: realistic conversations with inventors, examiners, partners, and clients 
• Supervision: reviewing AI-generated work product before anyone relies on it 
• Mentorship: the assignments and feedback that hold it together.

The purpose is the same across all of them: to build the strategic judgment AI cannot replace, deliberately in a setting where a mistake costs nothing, rather than leaving that judgment to accumulate by chance on live client matters. 

This article goes inside one of those modules, simulation, and walks through a single session the way a summer clerk or first-year associate experiences it, to show what the exercise is actually training. The first article in this series explained why the firm built SimProf. This one is more concrete. 

The SIMPROF setup 

A clerk opens a simulation scenario. The brief is short. The example we will use here is an inventor intake interview. Our firm tends to assign these early, and the scenario has a name that gives away the trap: the inventor says yes to everything. 

The clerk’s job is straightforward on the face of it. Interview the inventor, draw out the technical facts, and come away with enough to draft a sound set of claims. An AI plays the inventor and stays in character throughout, answering questions, reacting, volunteering some things and withholding others. 

The catch is the personality. An inventor who agrees with every suggestion produces a comfortable conversation and a dangerous record. Ask whether the device works at any temperature and an agreeable inventor says, sure, absolutely. You have learned nothing except that the inventor is agreeable. The failure mode the scenario is built around is the predictable one: the clerk writes broad claims the inventor’s actual prototype does not support. 

The decisions 

Inside the session, the clerk has to do the things that separate a useful interview from a pleasant one. Distinguish a preferred feature from a required one. Notice when an answer is enthusiasm rather than fact. Resolve an ambiguity with a follow-up question instead of an assumption. Confirm the fallback embodiments a durable specification will need. 

None of that is drafting. All of it decides whether the eventual draft is any good. Judgment comes before keystrokes, and the scenario is watching for it: whether the clerk surfaces the ambiguities and clears them, whether they separate what is preferred from what is required andwhether they lock down the positions a good claim set falls back on. 

The feedback

After the session, SimProf scores the work across four dimensions: technical accuracy, persuasiveness, strategy, and communication. The clerk sees where the interview went thin, which assumption went unchallenged, which single fact would have changed the claim scope. The debrief is the point of the whole exercise. 

A real inventor interview that goes badly costs a client a weaker patent, and the clerk may never learn why. A SimProf interview that goes badly costs nothing, and the clerk learns exactly why, immediately. That asymmetry is the entire argument for practicing this way.

“What I want a young attorney to take from an exercise like this is not a script. It is the instinct to be suspicious of an easy answer, to notice when they are being told what they want to hear. That instinct used to take years of live matters to build. We can build it deliberately now, and we can build it before there is a client on the other side of the table.”  — Piers Blewett, Principal, SLW

Why scenarios, not lectures

You can tell someone to validate an inventor’s facts. They will nod, and they will mean it. Reps are a different thing entirely: practicing the skill, getting it wrong somewhere safe, and feeling the consequence land. SimProf carries a library of these scenarios. The agreeable inventor is one. There is also the examiner who interrupts and anchors on an obvious-combination rejection, the partner who hands down contradictory edits under a deadline, the inventor who answers in three words and withholds everything else until asked the right way. Each one targets a specific judgment skill and a specific way of failing, and the difficulty climbs from beginner to expert as a trainee progresses. 

The simulator metaphor holds all the way down. We are not trying to replace the cockpit. We are trying to make sure that by the time a new attorney is in it, the hard moments are not the first time they have seen them. In the next article, Jeff Cobia and Aris Gregorian step back to show the full training apparatus SimProf sits inside. 

Learn more about SimProf.