How long does a 10-call analysis take?

Usually 2-4 minutes. Larger runs scale roughly linearly — per-call LLM analysis runs in parallel, so a 100-call run typically finishes in 7-10 minutes.

What's the difference between library prompts and writing my own?

Library prompts are curated by Discera for common use cases — closed-lost themes, objection patterns, competitive mentions. They're not editable. Writing your own gives full control over the deal-level prompt and an optional executive summary directive.

Can I re-run an old analysis with the same settings?

Yes. Open any past report and click Re-run. The settings (prompt, filters, date range) preload into `/new` so you can tweak and submit.

Why did my analysis return fewer calls than I asked for?

`max_calls` is a ceiling, not a target. If Gong + your filters yield fewer calls in the date range than the ceiling, Discera analyzes what's available and returns that count in the report.

Why does Max calls change when I pick a different date range?

It tracks the range you picked, so the run covers the whole window without you doing the arithmetic — up to a ceiling of 500 calls. If the window holds more than that, Discera tells you how many calls would be left out before the run starts. Type your own number any time and it stops adjusting.

How do I run my first analysis?

Getting started·May 31, 2026·2 min·By Discera Team

Run your first Discera analysis in five steps: open /new, pick or write a prompt, set max calls and date range, click Run, watch results.

TL;DR

Open /new, choose a prompt (library or your own), set max_calls (default 10 for first runs) and a date range (default last week), click Run, then watch the live progress at /runs/<jobId> and read the final report at /reports/<reportId>.

Steps

01Open /new. Discera shows a start-mode chooser: Choose from the library or Write your own prompt. Pick one.
02If you chose the library, pick a prompt — Discera ships a curated catalog with categories like closed-lost analysis, objection patterns, and competitive intel. The prompt and a suggested executive summary directive populate automatically.
03If you chose your own, write a deal-level prompt in plain English — one question Discera should answer for every call in scope. Optionally add an executive summary directive to focus the roll-up.
04Set the run scope: pick a date range (last week is the common default) and any HubSpot or Gong filters you want to narrow on. max_calls fills itself in with the number of calls in the range you picked, up to 500, and follows along each time you change the range — lower it if you only want a sample. Once you type your own number, Discera stops adjusting it.
05Click Run. Discera queues the analysis and redirects you to /runs/<jobId>, which streams progress with a heartbeat — Gong fetch, transcript pull, per-call analysis, executive summary, completion. When it finishes, the report opens at /reports/<reportId>.

Discera /new page showing the start-mode chooser with two cards: Choose from the library and Write your own prompt — Step 1: /new opens with a start-mode chooser — library prompt or your own.

What happens during a run

Discera fetches the calls matching your date range and filters from Gong, pulls each transcript, and runs your prompt against every call in parallel. The per-call findings feed into an executive summary that roll-ups themes, counts, and outliers across the whole batch. The finished report renders the executive summary, per-call detail, and a JSON payload you can download.

After the first run

If results look right, save the configuration as a preset from the report page — presets store the full prompt plus filter set, so future weeks reuse the same setup. To run the same prompt on a schedule, open /schedules and create a daily, weekly, or weekdays job. To share findings with the team, wire up Slack so completed runs post a notification automatically.