Your first research run.
From sign-in to a finished report in a few minutes. This walkthrough covers the web UI path: ask a question, watch the pipeline unfold, read the report, save sources to your library, and export citations in the format you need.
§ 01Sign up & sign in
Create an account at essarion.com and verify your email. Once verified, sign in to the workspace. ResearchAnything is available from the main navigation alongside Agents and the rest of the platform.
If you've been issued an API key already (an esk_ prefix), you can also drive ResearchAnything from the API — but for a first run, the web UI is the fastest path.
§ 02Run a query
Open ResearchAnything and type a question into the prompt. Treat it like a brief, not a search box. Specific, well-scoped questions produce better reports than broad ones.
What a good prompt looks like
- Specific. "What were the main findings of the 2024 IEA report on grid-scale battery storage?" beats "Tell me about batteries."
- Time-bounded. If you want recent material, say so. The recency score will respect it.
- Scoped. A single question per run; if you have three, run three.
Submit the run
Hit Run. The UI will switch to the live run view and start streaming events as the pipeline executes.
§ 03Watch the run unfold
The run view is a vertical timeline of phases, each one expanding as it runs and collapsing when complete. You'll see, in order:
- Analyzing query — the model classifies intent and extracts entities.
- Planning — one to three sub-queries are produced.
- Searching — search results stream in.
- Deduping & scraping — duplicates are removed and survivors are fetched.
- Screening & scoring — low-quality pages are dropped; survivors are scored.
- Deep analysis — claims, quotes, and evidence are extracted from the top sources.
- Citations & synthesis — citations are built and the report is written.
- Complete — the run is persisted and sources land in your library.
Each phase emits reasoning and structured output that you can expand inline. There is no hidden state — what you see is what was persisted.
§ 04Read the report
When the run completes, the view switches to the finished report. You'll see:
- An executive summary at the top — a single-page answer to the original question.
- The full report below it — sectioned, with evidence-backed claims and inline references.
- A sources panel on the side, showing the survivors and their scores.
- A citations panel, showing each source rendered in your preferred format.
§ 05Save sources to library
Sources from completed runs are saved automatically and idempotently to your library. You can then re-use them across other runs and projects without re-scraping.
To curate a subset, use the source panel's per-row save toggle, or save the entire run's sources from the run header. Saving twice is safe — the operation deduplicates by URL.
§ 06Export citations
Open the citations panel and pick your format — MLA, APA, Chicago Notes, Chicago Author-Date, IEEE, Harvard, or BibTeX. You can:
- Copy a single citation in the format you've selected.
- Copy all citations for the run as a single block.
- Download a bulk archive with citations in every format, plus a manifest.
For a deeper look at how citations are built and what fields they carry, see Citations.
§ 07Common gotchas
A query that's too broad
If the planner produces sub-queries that look unrelated to what you actually meant, your prompt was probably too broad. Re-run with a tighter version.
Paywalls and bot protection
Some publishers actively block scraping. Those sources will be dropped at the screening phase. The selection rationale will tell you what was excluded and why.
Recency vs authority tension
For fast-moving topics, the freshest material may not be the most authoritative. The composite score balances both, but if you want one or the other to dominate, say so in the prompt.
Daily limits
Hit the daily run cap and the start endpoint will return a quota error. The cap resets at the start of your billing day. Plan-level limits are visible in your account settings.