Evaluate before CRM. Your stack fires the outcome.
Semantic lead qualification for RevOps automation — a versioned rubric returns status, extracted_fields, missing_fields, and next_action. You branch HubSpot, Slack, or nurture. EchoStack does not act inside your CRM.
Evaluations run
— demo UI
Decision latency
~4s p50 target
Contract
status + fields
- [✓] budget: $50k+
- [✓] timeline: Q1
- [✓] authority: VP Sales
Caller branches on decision
next_action → book_meeting (caller)
Your n8n Switch → HubSpot owner
PARTIAL → ask missing_fields
The gap
Manual follow-up
vs. Evaluate-before-CRM.
Most teams lose momentum between signal and action. Drop one Evaluate step, then let your automation route while intent is still high.
The manual path
~2 days lost- 1
High-intent signal arrives
0mStrong buy signal lands in your stack.
- 2
Someone reviews it hours later
+2h"Maybe tomorrow."
- 3
Someone logs it in HubSpot
+1d"If they remember."
Follow-up goes out
+2d"Lead is already cold."
With echostack
< 5sQualified in seconds. status / missing_fields / next_action — your HubSpot or n8n workflow fires while the lead is still warm.
- Evaluated in seconds, not hours
- Caller updates CRM and calendar from the decision
- Right branch (owner, nurture, or Slack) while intent is still high
Decision returned
QUALIFIED · missing_fields[] · next_action
Evaluation substrate
Different criteria,
same decision contract.
Define what qualifies in plain language — BANT, MEDDIC, or custom detectors. The Evaluation API returns the same status / fields / next_action shape so your automation can branch reliably.
Sales frameworks like BANT are just examples. The product is the reusable layer underneath — criteria in, structured decision out, outcome downstream.
How it works
Set your rubric once.
Branch on the decision.
Create an Evaluation, call Evaluate with esk_ + evaluation_id, and let your automation Switch on status — including PARTIAL.
Step 01
Define what good looks like
Tell EchoStack what a qualified lead, urgent ticket, or ready-to-buy moment looks like — BANT, MEDDIC, or custom detectors in plain language.
- When a form answer set is QUALIFIED vs PARTIAL
- When a ticket should escalate before a human debates priority
- When a transcript is ready for the rep queue
Step 02
Wire Evaluate into your stack
Use an org API key (esk_…) and evaluation_id — or ad-hoc Criteria. Connect HubSpot, n8n, Zapier, or Make so your workflows act on the result.
- n8n community node + Switch on status
- Webhook normalize → POST /v1/evaluate
- Same contract from playground to production
Step 03
Turn it on — your stack routes
EchoStack returns the decision as signals arrive. Your automation books, updates CRM, nurtures, or alerts — while interest is still high.
- QUALIFIED → owner / stage in your CRM
- PARTIAL → ask for missing_fields
- FAILED → nurture or close the loop
What ships today
Lead with Evaluation API. Partners shape what comes next.
Evaluation API is GA — decisions only. Action Resolution and Full Platform are invite-only / design-partner shaped, not public GA SKUs.
Evaluation API
GA · decision only
Structured status, fields, and next_action. Your stack decides what fires.
Action Resolution API
Design partners · not GA
Auto CRM / calendar / Slack outcomes — shaped with partners; not sold as available today.
Full Platform
Design partners · not GA
Always-on voice, chat, and channels — invite-only; not a public GA package.
Integrations
Decide once.
Your tools act.
Wire Evaluate into the stack you already run. EchoStack returns the qualification decision; n8n, Zapier, Make, or your backend fire CRM updates, Slack alerts, and nurture.
CRM & revenue
Branch owner assignment and deal stage in HubSpot when Evaluate returns QUALIFIED — your workflow writes the CRM.
- HubSpot
Scheduling
On QUALIFIED, your automation books the meeting (e.g. Calendly) — EchoStack only recommends next_action.
- Calendly
Support
Score ticket text via support_triage Evaluate, then route in your helpdesk or n8n — no Zendesk-native agent.
- Ticket webhook
Notifications
Your Switch on status pings Slack or a channel — EchoStack does not post for you.
- Slack
Automation
Drop Evaluate before CRM or nurture. Companion to n8n, Zapier, and Make — not a replacement.
- n8n
- Make
- Zapier
Voice & AI agents
Call Evaluate as a tool hop from agents or post-call pipelines. Always-on voice channels are design-partner / Full Platform.
- LiveKit
- Voiceflow
- Retell
Webhooks & API
Send any signal via HTTP. Get structured decisions and outcomes back in JSON — wire into your own apps or internal tools.
REST · WebhooksEval API quickstartNot Zapier with an LLM.
Generic workflow tools chain steps. EchoStack owns the qualification decision — versioned rubrics, reusable Evaluations, and a stable status contract your automation branches on.
- Versioned rubric criteria — same rules every time
- Reusable Evaluations, not one-off GPT prompts
- Decision in seconds while the signal is still hot
Need something else? HTTP webhooks and REST work with any stack — caller normalizes the payload, then Evaluate.
Who it's for
One evaluate step.
Three ways in.
Same rubric-backed decisions everywhere — branch your automation, call our API from your agent, or qualify conversations as a design partner.
n8n · Make · Zapier
- Official n8n node + importable example workflow
- Eval-first: trigger → evaluate → Switch on QUALIFIED / PARTIAL / FAILED
- Same BANT/MEDDIC rules as API and HubSpot forms
REST · webhooks · your agents
- POST /v1/evaluate with org API key (esk_…) and evaluation_id
- Transcripts, forms, and ticket text — rubric resolved server-side
- Evaluation logs in the dashboard for audit and iteration
Calls · widgets · agent transcripts
- Post-call transcript scoring with the Evaluation API
- Playbook platform: voice, chat, and connectors (design partner)
- Mid-conversation evaluate + next_instruction for agents
Pricing
Simple pricing
for the decision layer.
Lead SKU is the Evaluation API — Free 500 evals/mo, then Scale at $0.02/eval. Action Resolution and Full Platform stay partner-shaped until GA.
What ships today
Lead with Evaluation API. Partners shape what comes next.
Evaluation API is GA — decisions only. Action Resolution and Full Platform are invite-only / design-partner shaped, not public GA SKUs.
- One rubric — reused across forms, webhooks, and transcripts
- Caller-owned outcomes — CRM, calendar, Slack via your automation
- Roadmap surfaces — Action Resolution and Full Platform with partners
Same rubrics at every tier — you change who pulls the trigger, not the criteria.
Qualification decision API · GA
Free
500 evaluations / month · then Scale at $0.02 / eval
- lead_qualification, support_triage, intent_detect, or custom
- Structured JSON: status, fields, missing_fields, next_action
- n8n node + webhook guides — caller fires CRM / Slack / nurture
Design partners · not GA
Invite only
Partner-shaped — not a public GA SKU
- Influence CRM / calendar / Slack auto-outcome priorities
- Not available for self-serve checkout
Design partners · not GA
Invite only
Always-on channels — partner access only
- Voice, chat, and webhook channels under partner setup
- Not a public Custom checkout SKU today
An Evaluation is one versioned rubric (e.g. inbound BANT). Scale meters Evaluate usage. Action Resolution outcome fees apply only when that product ships and executes an outcome for you.
Live sandbox
Signal in. Structured decision out.
Paste a transcript, form, email, or ticket and run a live evaluation. See qualification status, extracted fields, and a recommended next action — then wire the same contract in production.
Try a scenario
Interactive playground — no account required
Pick from sales call, web form, support ticket, or trial usage scenarios. Same Evaluate engine as production.
Design partners
Workflows partners
are shipping.
We only publish metrics and quotes with written customer approval. These are the production workflows design partners run today.
Inbound leads
Form or webhook → BANT Evaluate in seconds → your CRM or Slack workflow fires on status.
Post-call transcripts
Transcript → structured fields + next_action → rep queue without manual note review.
Support triage
Ticket text → severity Evaluate → escalate or queue before a human debates priority.
Design partner program
Ship one Evaluate-before-CRM workflow.
Shape what we build next.
We're working with RevOps and automation builders who have one inbound form or webhook stuck in manual review or a GPT step. You get white-glove Evaluation setup; we need honest production feedback for 4–6 weeks.
Who
- RevOps, sales ops, or automation owner (~20–200 person B2B)
- Named inbound form/webhook + BANT/MEDDIC/custom rubric
- Technical owner who can wire esk_ + evaluation_id in n8n, HubSpot, or HTTP this month
You get
- White-glove lead_qualification Evaluation + org API key (esk_…)
- Weekly 30-min check-in with the founder
- Early influence on Action Resolution (auto-outcomes) — not sold as GA
- Input on packaging before public launch
We need
- Evaluation API live on real traffic ≥4 weeks (not sandbox only)
- Honest weekly feedback
- Permission for an anonymized case study when you're ready — no invented metrics
Close the gap between signal and decision.
Define your rubric once. Call Evaluate. Let your stack fire CRM, Slack, or nurture while intent is still high.