The principle that claims are not necessarily limited to an embodiment described in the specification is well established. That cannon is not immutable, and unrecited features have been found to nevertheless limit a claim when, for example, the specification disparages prior art solutions that lack the feature and when the specification describes that feature as being part of “the invention.” However, even those circumstances may not be enough, as evidenced by the Federal Circuit’s recent decision in Applications in Internet Time, LLC v. Salesforce, Inc., Appeal No. 2024-1133 (Fed. Cir. Oct. 10, 2024).
At issue was whether the claim term “automatically detecting changes,” as recited in claim 1 of U.S. Patent No. 7,356,482 (“the ’482 patent), required the use of “intelligent agents.” The district court found that the ’482 patent expressly disavowed automatically detecting agents without intelligent agents and thus construed the term narrowly to require intelligent agents. The district court relied on the ’482 patent’s identification of intelligent agents in connection with describing “the invention.” For example, the ’482 patent states that “[t]he invention begins tracking changes using one or more intelligent agents” and that “FIG. 2 is a flowchart illustrating use of the invention to respond to one or more relevant changes found by an intelligent agent on a network.” The ’482 patent also states that “[t]he system operates at four layers, as illustrated in FIG. 1,” and that figure shows the use of “intelligent agents” at the first layer:
The district court also relied on the ’482 patent’s discussion of the prior art. Specifically, the ’482 patent noted that a limitation of prior solutions was that they did not “provide[] a comprehensive, integrated framework for . . . automatically making application and database changes using intelligent agent routines.”
Under this construction, the district court granted summary judgment of non-infringement. The Federal Circuit reversed.
The Federal Circuit focused on two considerations in arriving at a contrary construction. First, invoking the doctrine of claim differentiation, the Federal Circuit noted that unasserted claim 8, which depends from claim 1, recites “wherein the change management layer further comprises one or more intelligent agents that detect changes that affect an application.” According to the Federal Circuit, the presence of “intelligent agents” in dependent claim 8 is a “strong reason” not to limit independent claim 1 to require intelligent agents.
Second, the Federal Circuit found that the specification’s discussion of “intelligent agents” was merely “exemplary and not descriptive of the invention as a whole.” The Federal Circuit found important that the Abstract, Background, and Summary of the Invention do not discuss intelligent agents in connection with the solution of the patent. With respect to the criticism of the prior art failing to use intelligent agents, it did not rise to the level of clear disavowal.
The Federal Circuit thus modified the district court’s construction and vacated the grant of summary judgment of non-infringement.