SynthChat

SynthChat is an assistant that can run built-in commands (e.g. detect AI, check readability, summarize) from the conversation. Ask in plain language and it will use the right tool and show you the result. Useful for quick checks or when you prefer a chat over opening each tool separately.

SynthChat is a conversational control layer on top of SynthQuery’s existing tools. Instead of navigating to each page, you describe intent in natural language—summarize this policy, check this paragraph for AI-like sentences, run readability on my draft—and the assistant routes the request to the same server-side capabilities as the standalone apps: detection, SynthRead, summarization, grammar, plagiarism, humanizer, and others enabled for your account. Responses appear in the thread with structured results where the product supports them. The assistant may confirm which tool ran so you avoid silent misroutes. Chaining is possible in principle (detect, then suggest edits) when policies and safety settings allow, but you should still review outputs critically—chat does not replace the detailed visualizations on dedicated pages such as sentence heatmaps. SynthChat shares the same models and limits as direct tool usage; it is not an unlimited parallel product. Enterprise admins may restrict certain commands. Treat transcripts according to your data policy when content is sensitive.

Busy editors orchestrate multi-step QA in one session: summarize a long brief, check readability, then scan for similarity—pasting context once. Students on allowed workflows ask for explanations of scores alongside the underlying tool output. Support leads prototype macros through chat before saving templates. Marketing managers run ad-hoc checks while brainstorming copy in-thread. Remote teams collaborate asynchronously when chat history is permitted, logging decisions next to results. When fine-grained controls matter—heatmap inspection, glossary editing, precise plagiarism match review—users still open the dedicated tool pages for full UI fidelity.

State your goal plainly: audience, constraints, and which artifact you are checking. Paste text or provide URLs only when the product supports fetching; if extraction fails, paste manually. Ask explicitly for the tool you need if the assistant seems uncertain—clarity reduces misroutes. After each result, open the dedicated tool page when you need heatmaps, match lists, or fine settings. Log or export transcripts if compliance requires audit trails. Trim sensitive identifiers before pasting when policies demand minimization. Chain steps only when each prior output has been reviewed—do not humanize undisclosed AI text if your institution forbids it. Monitor usage dashboards: chat convenience can consume quota quickly on large pastes. For automation at scale, prefer APIs; use chat for human-in-the-loop steering.

Chat may summarize or paraphrase tool output—verify against raw results on specialist pages for high-stakes decisions. Transcripts may be stored; avoid secrets in plain text when policy forbids it. Admins can disable tools—unsupported requests should fail clearly. Do not treat chat as legal advice. Pair conversational speed with human judgment and institutional policy.