Monitoring

A lot of patent research has to be repeated. The patent offices publish new applications and grants every week, competitors keep filing and a landscape you once mapped keeps growing. Keeping up by hand means running the same searches again and again and working out each time what is new.

For that you can put a research agent on a schedule. The agent then repeats its task at the times you set, and every run builds on what the earlier runs found.

Watching patents#

For patents we provide three standard skills that set up a weekly watch for you:

• Monitor Competitor follows one or more companies. The agent first works out every name a company files under, such as its subsidiaries and former names, and checks the list with you.
• Monitor Landscape follows a technology, which the agent defines together with you by classes and key terms.
• Monitor Classifications follows a set of CPC classes.

You pick the skill when you start a new conversation, or simply ask the agent to use it. The agent also asks what you want to learn from the watch, e.g. which of your own products or patents it should compare new filings against.

What a watch does#

The agent first adds a baseline to your research, e.g. everything the companies published over the last year, so that the first run only reports what is new. It also creates a note for the watch, with an overview at the top and one dated section per run.

Each run then adds the patents that were published since the last run. The agent reads every new patent, sums up in one line what it protects and puts first what touches the products or patents you named. It also points out what changed, e.g. an application that was granted or a company that started filing in a new territory.

Every patent gets the day it entered the watch in a date attribute named First seen. To see what a run added, simply filter the patents table by it.

Watching products#

New products keep reaching the market, and you find them with a deep web search. Just ask the agent to repeat such a search every day, week or month. Each repeat only collects products it has not found before, and the agent adds the ones that hold up to your research. The Infringement Detection skill offers this as its last step and also charts the new products against your patents.

Be aware that a repeating search keeps running, and keeps using your organization's AI & Data usage, until you ask the agent to stop it.

Schedules#

The watches above are built on schedules, and you can put any other work on a schedule as well by simply asking the agent, e.g. "Look for new patents on solid-state batteries every Monday at 9:00 and tell me what changed". A schedule can run once or repeat every few hours, days, weeks or months.

A card in the conversation shows when the schedule runs next, and the sidebar lists your scheduled agents. There are no buttons for schedules. To change, pause, resume or delete one, you just ask the agent in the same conversation.

Reports and e-mails#

Since you are usually not around when a run happens, the agent settles as much as it can on its own. It e-mails you the result when there is something you should see, and a question when it can't go on without a decision from you. You can choose which e-mails you want under Settings > Notifications.

When a run fails#

If the runs of a schedule keep failing, e.g. because your organization's AI & Data usage has run out, the schedule pauses and we e-mail you. Once you have fixed the cause, open the conversation and simply ask the agent to resume it.

Costs#

You pay for what scheduled runs use, just like for any other agent run. You can find all rates on our pricing page.