| name | sales-trill |
| description | Trill (trill.so) platform help — a free, public-beta AI qualitative-research assistant that turns interview transcripts into insights: import interviews and its AI auto-generates observations, detects and categorizes themes across sessions, and (via a Notion-like editor) drafts stakeholder reports. It is the analysis end of the research stack, a lighter peer of Notably, Dovetail, Condens, and Aurelius. Use when synthesizing user interviews into themes without manual coding, analyzing qualitative interview data, drafting a research report from interviews, deciding whether Trill fits a UX/market/academic research workflow, choosing between Trill and Notably or Dovetail, or getting data out when there is no public API. Do NOT use for choosing a qualitative-analysis tool across the whole market (use /sales-customer-feedback) or recruiting and running studies end-to-end (use /sales-great-question). |
| argument-hint | [describe what you need help with in Trill] |
| license | MIT |
| version | 1.0.0 |
| tags | ["sales","customer-cx","platform"] |
Trill Platform Help
Trill (trill.so) is a free, public-beta AI qualitative-research assistant — the analysis end of the research stack and a lighter peer of Notably, Dovetail, Condens, and Aurelius. You already have the data (interview recordings/transcripts); Trill ingests it, auto-generates insights/observations aligned to your project objectives, detects and categorizes themes across interviews so you skip manual line-by-line coding, and drafts stakeholder reports in a Notion-like editor.
Two things to say almost every time:
- Trill ANALYZES interviews — it does not recruit, schedule, or run them. No participant panel, no moderation, no data generation. If the user needs to find and talk to users, that's a research-ops job — route to
/sales-great-question (recruit + run) or /sales-idea-validation / /sales-chikka (AI-moderated interviews that generate the data).
- Trill has NO public API, webhooks, MCP server, or documented Zapier. It is a free public-beta UI tool; data-out is manual. A "pipe insights into my CRM/warehouse" ask is a manual-export job, not an endpoint call — route a documented pipeline to
/sales-userintuition.
Step 1 — Gather context
If references/learnings.md exists, read it first for accumulated platform knowledge.
Ask only what you can't infer:
- What do you want from Trill?
- A) Import interviews — bring in transcripts/recordings to analyze
- B) Generate insights — AI observations aligned to project objectives
- C) Theme detection — auto-cluster/categorize themes across interviews (skip manual coding)
- D) Report — draft a stakeholder report in the Notion-like editor
- E) Get data out — manual export (no API)
- F) Compare / decide — Trill vs Notably/Dovetail/Condens, or whether it fits your workflow
- Do you already have interview data, or do you still need to recruit/run the interviews? The second is a research-ops question — flag it in Step 2 and route out.
Skip-ahead: if the prompt already names the task or the question is specific, go to Step 3.
Step 2 — Route or answer directly
| If the user's question is about… | Route to |
|---|
| Choosing a qualitative-analysis / research tool across the market, or VoC/feedback strategy | /sales-customer-feedback {question} |
| Recruiting + running studies end-to-end (panel, scheduling, methods) then storing them | /sales-great-question {question} |
| Validating whether to build an idea (evidence ladder, AI scores, smoke tests) | /sales-idea-validation {question} |
| Running AI-moderated interviews to generate the data first | /sales-chikka or /sales-idea-validation {question} |
| A documented API / webhook-native research pipeline (Trill has none) | /sales-userintuition {question} |
| The most AI-forward repository peer (auto theme extraction + insight summaries, canvas) | /sales-notably {question} |
When routing, give the exact command: "This is a {domain} question — run: /sales-great-question {original question}"
Otherwise, answer Trill-specific questions using Step 3.
Step 3 — Trill platform reference
Read references/platform-guide.md for the full reference — the capability/automation-surface table (what's UI-only), the free public-beta status and what "no pricing yet" means for your workflow, the Project → interview → insight → theme → report data model, the import workflow, and the no-API data-out playbook. Trill has no REST API, so there is no API-reference file.
Answer using only the relevant section — don't dump the full reference.
Step 4 — Actionable guidance
- In every response, treat AI insights and themes as a draft, not a finding — verify against the source verbatim. Trill's auto-generated observations and theme categories are a starting draft; tell the user to open the underlying interview quotes and confirm the AI read them correctly before presenting an insight to stakeholders. Say this even when the user only asked how to speed up synthesis — AI acceleration is the value, but an unverified AI theme can misrepresent what a participant actually said.
- Draw the analysis-only boundary head-on whenever recruiting/data-collection comes up. Trill does not recruit participants, schedule sessions, or run/moderate interviews — it only analyzes interviews you already have. If the user has no interview data yet, say so plainly and route the recruit-and-run job to
/sales-great-question, or the AI-moderated interview-generation job to /sales-chikka or /sales-idea-validation.
- Flag it as export-only whenever any automation/API/pipeline comes up. Trill has no public REST API, webhooks, MCP server, or documented Zapier (verify — a beta may add them). The only data-out is manual export/copy, so "sync insights to HubSpot/Snowflake on a schedule" is a manual job, not REST — if a documented pipeline is required, route to
/sales-userintuition.
- Present status and pricing as best-effort: it is a free public beta, so plan for churn. State plainly that Trill is free while in public beta with no published paid tiers — which means features, limits, and data-retention can change without notice and a paid plan may appear later. Point the user to trill.so to confirm current status, and advise keeping the original transcripts as the source of truth rather than relying on Trill as the only store.
- Match Trill to the moment in the workflow and name its peers. Recommend it for the synthesis step — import the interview transcripts and let Trill auto-generate insights and detect/categorize themes across interviews so the user skips manual line-by-line coding, then draft a shareable report. Position it as a lighter, free option next to Notably (most AI-forward: theme extraction + insight summaries on a data canvas), Dovetail/Condens (heavier repositories), and Aurelius; for choosing across that market, route to
/sales-customer-feedback.
If you discover a gotcha or tip not in references/learnings.md, append it there with today's date.
Gotchas
Best-effort from research (2026-07) — status, feature set, and the no-API surface; the live site is a JS-rendered SPA (root 404s to fetchers) so specifics lean on third-party sources — verify at trill.so.
- Wrong-domain trap: the product is
trill.so, NOT trilliai.com. trilliai.com is an unrelated Windows VPS host; "Trill" also collides with Microsoft's open-source Trill streaming engine and the "Trill Project" social app. Confirm you're pointing users to trill.so.
- Analysis-only — no recruiting, scheduling, moderation, or data generation. Trill starts after you have interviews. For recruit → run → repository, use
/sales-great-question; for AI-moderated interviews, /sales-chikka.
- No REST API, no webhooks, no MCP, no Zapier. Data-out is manual export/copy only. Don't design a live pipeline around it — route pipeline needs to
/sales-userintuition.
- AI insights/themes are a draft, not a finding. Auto-generated observations and theme categories can misread nuance — always confirm against the underlying interview verbatims before presenting.
- Free public beta = expect change. No published paid tiers; features, limits, and retention can shift and a paid plan may arrive. Keep original transcripts as the source of truth.
Related skills
/sales-customer-feedback — The tool-agnostic qualitative-analysis / VoC strategy and the full research-repository landscape (use this to choose an analysis tool across the market, or for NPS/CSAT/survey strategy). Install: npx skills add sales-skills/sales --skill sales-customer-feedback -a claude-code/sales-notably — The most AI-forward research-repository peer — auto theme extraction + insight summaries on a split-screen data canvas, searchable cross-project repository; also no public API. The heavier, more feature-complete alternative to Trill's free beta. Install: npx skills add sales-skills/sales --skill sales-notably -a claude-code/sales-great-question — The all-in-one research-ops peer that recruits and runs studies (panel + scheduling + every method) then stores them — the pick when you have no interview data yet. Install: npx skills add sales-skills/sales --skill sales-great-question -a claude-code/sales-chikka — An AI voice interviewer that collects the qualitative data conversationally (multilingual, via a share link) — pair it upstream of Trill when you need to generate interviews first. Install: npx skills add sales-skills/sales --skill sales-chikka -a claude-code/sales-idea-validation — The validate-before-building method + the full research/validator landscape; where a research-synthesis tool fits among validators and interview tools (synthesis explains the why but is not demand). Install: npx skills add sales-skills/sales --skill sales-idea-validation -a claude-code/sales-userintuition — The API/webhook/MCP-native research peer — the pick when you need a documented programmatic pipeline Trill can't offer. Install: npx skills add sales-skills/sales --skill sales-userintuition -a claude-code/sales-do — Not sure which skill to use? The router matches any sales objective to the right skill. Install: npx skills add sales-skills/sales --skill sales-do -a claude-code
Examples
Example 1: "I ran 10 user interviews — how do I get themes without coding every line?"
User says: "I've got a stack of interview transcripts and I don't want to tag every sentence by hand. I want the main themes and a summary I can share."
Skill does: Recommends importing the transcripts into a Project, letting Trill auto-generate insights aligned to the project objective and detect/categorize themes across the interviews, then drafting a summary in the Notion-like editor. Insists the user open the underlying quotes under each theme to confirm the AI read them correctly before presenting, and notes it's a free public beta so keep the original transcripts as the source of truth.
Result: The user gets a verified, shareable set of themes in a fraction of manual-coding time.
Example 2: "Can I pull Trill insights into our warehouse / automate the export?" (developer/automation)
User says: "I want research insights flowing into Snowflake automatically instead of copy-pasting."
Skill does: States plainly that Trill has no public REST API, webhooks, or MCP server — the only data-out is manual export/copy. Suggests a manual export → ETL workaround, notes there's no programmatic pull so it's batch not a live sync, and — if a documented REST + webhooks pipeline is a hard requirement — routes to /sales-userintuition.
Result: The user builds a manual-export workflow instead of hunting for an API that doesn't exist.
Example 3: "Should I use Trill or Notably — and haven't recruited anyone yet"
User says: "I'm picking a research tool. We don't have interviews yet. Is Trill basically Notably?"
Skill does: Clarifies Trill is analysis-only (a lighter, free-beta peer of Notably that starts after you have interviews), and that Notably is the more feature-complete / most AI-forward repository. Because the user hasn't recruited yet, flags that neither is the right first step and routes the recruit → run job to /sales-great-question (or /sales-chikka for AI-moderated interviews); for choosing across the market, routes to /sales-customer-feedback.
Result: The user picks the right tool for their stage — collect data first, synthesize (Trill/Notably) after.
Troubleshooting
"The AI theme doesn't match what the participant actually said."
Symptom: An auto-generated theme or insight reads as off, generic, or overstated.
Solution: Treat AI output as a draft, not a finding. Open the interview quotes underneath the theme and confirm Trill clustered them correctly — re-categorize or split the theme where it conflated distinct points. The value is acceleration; the accuracy check is yours. Feed cleaner, well-scoped transcripts for better clustering.
"How do I get my insights out — and can I automate it?"
Symptom: The user wants results flowing into other tools.
Solution: In: import interview transcripts/recordings into a Project. Out: manual export/copy only — Trill has no public REST API, webhooks, or MCP, so there's no programmatic pull and no live sync. Do a manual export into your own ETL if you must, and route any documented-pipeline requirement to /sales-userintuition.
"Is Trill going to stay free — can I rely on it long-term?"
Symptom: The user wants to standardize a team workflow on Trill.
Solution: Trill is a free public beta with no published paid tiers, so features, limits, retention, and pricing can change without notice and a paid plan may appear. Confirm current status at trill.so, keep original transcripts as the source of truth (don't let Trill be the only store), and for a more established/feature-complete option compare Notably, Dovetail, or Condens via /sales-customer-feedback.
