DocsSearch Guide / Trends & portfolio

Trends & portfolio analysis

Map the market and competitive landscape via applicant ranking, yearly trends, and IPC distribution.

When to use this

This is the opposite of deep-reading a single document — you aggregate thousands of filings at once to see the big picture. Used for client reports, technology-trend write-ups, R&D planning, M&A due diligence, etc.

FindIP exposes a dedicated search_with_stats tool for this. You can group by year, country, applicant, or IPC in a single call, which makes it trivial for the LLM to render tables.

Recommended LLM prompt patterns

Use the patterns below as-is, or swap in your own variables. Each prompt is paired with the FindIP tool the LLM will call internally.

1. Baseline — trend + top applicants

Prompt

"For the noise-correction algorithm field of autonomous-driving LiDAR sensors, give me a table of yearly filing trends across US · KR · JP · CN · EP for the past 10 years, plus the top 10 applicants."

Tools: search_with_stats(stats_group_by=["year","applicant","country"])

2. Deep-dive on one applicant's portfolio

Prompt

"For Samsung Electronics' KR / US filings on OLED burn-in correction over the past 5 years, summarize the IPC distribution and yearly trend. Include title and claim 1 for the 5 core rights."

Tools: applicant_resolve("Samsung Electronics")search_with_stats(applicant_entity_ids=[...], stats_group_by=["year","ipc"])search_patents(...)

3. Competitor comparison matrix

Prompt

"In the all-solid-state battery field, build a comparison matrix of yearly filing trends and IPC distribution across LG Energy Solution, SAMSUNG SDI, CATL, and Toyota."

Tools: applicant_resolve(each company)search_with_stats(applicant_entity_ids=[...union of 4...], stats_group_by=["year","applicant","ipc"])

4. Surface emerging entrants

Prompt

"Among applicants whose filings have surged in the past 3 years in SiC power semiconductors, pick 5 newcomers that didn't exist 5 years ago. Include 2 key patents per company."

Tools: search_with_stats(stats_group_by=["year","applicant"]) → diff analysis

5. Detect shifts in technology focus

Prompt

"Tell me how the IPC distribution in this field has changed vs 5 years ago. Separate IPCs that are newly rising from those in decline."

Tools: search_with_stats(stats_group_by=["year","ipc"]) (compare 5-year windows before / after)

Reading the results

  1. Instead of taking raw stats as plain text, explicitly ask for "a table" — the LLM will lay it out in a much more readable form.

  2. For applicant comparisons, always go through applicant_resolve first. "Samsung Electronics" is registered under different name forms in KR / US / JP, and a naive search misses a lot.

  3. Filing volume ≠ technical strength. Ask the LLM to also surface secondary signals like citation count or triadic-patent rate (KR + US + JP simultaneous filings).

FindIP — Semantic Patent Search