TextRazor

@TextRazor extract all named entities from this press release and tell me which companies, people, and locations are mentioned

TextRazor identified 12 named entities in the text: 3 organizations (Acme Corp, TechVentures LLC, DataFlow Systems), 4 people (Sarah Chen as CEO, Michael Torres as CTO, Dr. Emily Rodriguez, James Park), 3 locations (San Francisco headquarters, Austin office, Singapore), and 2 products (CloudSync Platform, API Gateway). Each entity includes confidence scores ranging from 0.87 to 0.98 and links to knowledge-base identifiers where available.

This example shows TextRazor's core entity extraction strength — identifying and categorizing named entities with confidence scores. The API requires an API key and processes text synchronously, so very long documents (over 200KB) may hit rate limits. Entity linking to knowledge bases depends on the entities being well-known in public datasets.

@TextRazor analyze the dependency trees in this customer feedback to understand the grammatical structure and identify what subjects are connected to negative sentiment words

TextRazor parsed 8 sentences into dependency trees, revealing grammatical relationships. In sentence 3, 'interface' (subject) connects via nsubj relation to 'confuses' (verb), with 'users' as the direct object. Sentence 5 shows 'checkout' (subject) linked to 'fails' through a similar structure. The analysis identified 14 verb-object pairs and 22 modifier relationships, making it clear which product features are grammatically tied to complaint verbs versus praise verbs.

Dependency tree analysis is powerful for understanding sentence structure beyond keyword matching — useful when sentiment analysis needs grammatical context. This example pairs TextRazor's syntactic output with your own interpretation layer. Note that complex or fragmented sentences may produce ambiguous trees, and the tool works best with well-formed English text.

@TextRazor create a custom entity dictionary called 'internal-products' with our product codenames so future analyses recognize them as entities, then classify this support ticket

Dictionary 'internal-products' created with 7 entries (Project Falcon, Nighthawk API, Redwood Dashboard, etc.). TextRazor will now recognize these as custom entities in subsequent analyses. Classifying the support ticket: primary category 'Technical Support / API Issues' (confidence 0.91), secondary category 'Billing Questions' (confidence 0.34). The ticket mentions 'Nighthawk API' (now recognized as a custom entity) 4 times, suggesting it's the focus of the user's concern.

This example demonstrates TextRazor's dictionary management for domain-specific entity recognition — critical when analyzing internal documents with proprietary terminology. Custom dictionaries persist across API calls and require the dictionary management permission. Classification categories come from TextRazor's pre-trained models; you cannot define custom categories without their enterprise tier.

When TextRazor wins for support teams drowning in unstructured feedback

A 6-person support team at a B2B SaaS company gets 200+ customer emails daily, plus Slack messages and survey responses. They need to route feedback to product, engineering, or account management without reading every word. TextRazor's classification and entity extraction tools tag each message with product areas, sentiment, and named features in under a second. The team sets up a shared Switchy workspace where the MCP runs on incoming feedback, auto-tags tickets in Linear, and surfaces trending issues in a daily digest. This works when your feedback volume justifies the API cost (roughly $0.50 per 1,000 analyses) and you can define 8-12 stable categories. If your product changes weekly or you get fewer than 50 messages a day, manual tagging is faster. For teams at 100+ daily inputs with consistent taxonomy, TextRazor pays for itself in saved triage time.

How TextRazor catches risky language in marketing copy before publication

A 4-person marketing team at a fintech startup writes blog posts, email campaigns, and landing pages that must avoid specific claims (like 'guaranteed returns' or unapproved medical terms). They use TextRazor's dependency tree and entailment extraction to flag sentences that imply prohibited statements, even when the exact phrase isn't present. The MCP runs in a Switchy workspace connected to their Google Docs drafts, highlighting risky constructions before legal review. This setup works when you have a defined list of 20-50 prohibited concepts and publish 10+ pieces per month. The dependency tree analysis is overkill for simple keyword blocking—use a regex if you're just banning exact phrases. For regulated teams that need semantic compliance checks without a full-time compliance officer, TextRazor's linguistic depth justifies the API key and 13-tool learning curve.

When TextRazor's entity linking beats manual note-taking for market research

A 3-person strategy team at a Series A company monitors 15 public competitors by reading quarterly earnings transcripts. They need to track which products, partnerships, and geographies each competitor mentions, then compare trends over time. TextRazor's named entity extraction and phrase tools pull structured data from unstructured transcripts—company names, product mentions, geographic markets—and link entities to Wikidata for disambiguation. The team drops transcripts into a Switchy workspace, runs the MCP, and exports a CSV of entities by quarter. This works when you're analyzing 5+ long documents per month and need consistent entity tagging across sources. If you're only tracking 2-3 competitors or reading summaries instead of full transcripts, a spreadsheet and manual highlights are faster. For research teams building longitudinal datasets from public filings, TextRazor's entity linking saves 4-6 hours per analyst per week.