Now in early access

Catch failing support replies before they become complaints.

Formend scores every support conversation in real time — detecting missed intent, flagging failures, and routing the right alerts to the right supervisors.

Request early access See how it works

Works with Zendesk, Intercom, Freshdesk, Slack and custom webhooks.

conv_001 · agent_42 · 10:44 UTC · sample dataEscalated

Customer message

"I've been waiting 3 weeks and nobody is helping me. This is completely unacceptable."

Agent reply

"Sorry for the inconvenience. Please check our help centre for updates on your order."

Conversation score

0/ 100

Resolution risk
Policy breach detected

⚑ Resolution risk flagged

⚑ Tone policy breach

Policy signals 3 detected

Escalation tier Supervisor

Configurable rules Active

Configurable policy engine

#support-alerts — Supervisor alerted · conv_001 · Resolution risk detected

chats per agent
per day

100

supervisors can
monitor manually

Real time

scoring — not
monthly reviews

Custom

configurable
escalation rules

The problem

Scripts get sent.
Flags get ignored.
Reports arrive sanitised.

By the time a pattern surfaces in your QA review, the damage is already done.

01 / Agent failure

Agents default to scripts instead of resolving the issue.

Customers feel unheard. The same complaint returns.

02 / Supervisor failure

Flags are marked resolved without the conversation being read.

Accountability is theatre. Problems compound in silence.

03 / Reporting failure

Agencies self-report performance to the companies paying them.

The data is clean. The customer experience is not.

The solution

01 / Analyse

Read the customer's intent before the agent replies.

Formend surfaces urgency, emotion, and required action from every incoming message — before the conversation goes wrong.

Intent extractionSample data

"I've been waiting 3 weeks and nobody is helping me. This is completely unacceptable."

Intent detectedDelivery escalation — immediate resolution required

Urgency: HighSentiment: FrustratedAction: Immediate resolution

02 / Evaluate

Compare the reply against what the customer needed.

Each response is checked against configurable quality signals and policy rules. Failures are identified the moment the reply is sent.

Reply assessmentSample data

"Sorry for the inconvenience. Please check our help centre for updates."

Configurable risk threshold not met

Policy breach detectedResolution riskTone mismatch

Evaluation model · signal weights · 4 policy rules active · escalation tier: supervisor ········

03 / Escalate

Route failures to the right supervisor — with full context.

When a conversation falls below threshold, Formend alerts instantly. Supervisors see what happened and why, not just a ping.

Supervisor alertedSample data

# support-alerts

Formend · just now
conv_003 · Resolution risk detected. Customer intent unresolved. Supervisor action required.

Resolution riskPolicy breachEscalation logged

For developers

Three endpoints.
Works with your existing support stack.

Point Zendesk, Intercom, Freshdesk, Slack, or any webhook-enabled support tool at Formend. Add your API key. You're live in around 30 minutes.

Time to first integration: ~30 minutes

POST/v1/analyse— Intent + urgency

POST/v1/evaluate— Reply scoring

POST/v1/flag— Supervisor alert

View full API docs →

# POST /v1/evaluate

curl -X POST https://api.formend.tech/v1/evaluate \
  -H "Authorization: Bearer FORMEND_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "conversation_id": "conv_001",
  "customer_message": "I've been waiting 3 weeks",
  "agent_reply": "Please check our help centre."
}'

# Response
{
  "risk": "high",
  "outcome": "supervisor_alerted",
  "reasons": ["resolution_risk", "policy_breach"]
}
// POST /v1/evaluate

const response = await fetch('https://api.formend.tech/v1/evaluate', {
  method: 'POST',
  headers: {
    'Authorization': `Bearer ${FORMEND_API_KEY}`,
    'Content-Type': 'application/json'
  }{,
  body: JSON.stringify({
    conversation_id: 'conv_001',
    customer_message: 'I\'ve been waiting 3 weeks',
    agent_reply: 'Please check our help centre.'
  })
});

const { risk, outcome, reasons } = await response.json();
# POST /v1/evaluate

import requests

response = requests.post(
  "https://api.formend.tech/v1/evaluate",
  headers={"Authorization": f"Bearer {FORMEND_API_KEY}"},
  json={
    "conversation_id": "conv_001",
    "customer_message": "I've been waiting 3 weeks",
    "agent_reply": "Please check our help centre."
  }
)

data = response.json()
# data["outcome"] == "supervisor_alerted"
# POST /v1/evaluate

require 'net/http'
require 'json'

uri  = URI('https://api.formend.tech/v1/evaluate')
req  = Net::HTTP::Post.new(uri)
req['Authorization'] = "Bearer #{FORMEND_API_KEY}"
req['Content-Type']  = 'application/json'
req.body = {
  conversation_id:  'conv_001',
  customer_message: 'I\'ve been waiting',
  agent_reply:      'Please check our help centre.'
}.to_json

http         = Net::HTTP.new(uri.hostname, 443)
http.use_ssl = true
data         = JSON.parse(http.request(req).body)
// POST /v1/evaluate

$ch = curl_init('https://api.formend.tech/v1/evaluate');
curl_setopt_array($ch, [
  CURLOPT_POST           => true,
  CURLOPT_RETURNTRANSFER => true,
  CURLOPT_HTTPHEADER     => ['Authorization: Bearer ' . $FORMEND_API_KEY],
  CURLOPT_POSTFIELDS     => json_encode([
    'conversation_id'  => 'conv_001',
    'customer_message' => 'I\'ve been waiting',
    'agent_reply'      => 'Please check our help centre.'
  ])
]);
$data = json_decode(curl_exec($ch), true);

Works with helpdesks, CRMs, chat tools, and any outbound webhook.

Zendesk

Intercom

HubSpot

Slack

Salesforce

Gmail

Shopify

Freshdesk

Webhooks

Zendesk

Intercom

HubSpot

Slack

Salesforce

Gmail

Shopify

Freshdesk

Webhooks

Pricing

Only pay for conversations analysed.

Start small. Monitor everything. Scale as your support volume grows.

Starter

3,000 analyses / mo

£99

per month

£0.02 per analysis after limit

Intent analysis & scoring
Mismatch detection
Slack & email alerts
Basic audit logs
Webhook integration

Built for accountable support operations.

Formend is in early access. We build transparently and honestly — no certifications claimed before they are real.

Audit logs for everything

Every score, flag, and escalation is permanently logged with timestamps and agent IDs. Full traceability.

GDPR-aware workflow design

Customer data is used for evaluation only, not model training. Data retention controls available on request.

Role-based access controls

Agent, supervisor, and parent company access tiers planned. Each level sees only what they need to see.

API key authentication

Every endpoint requires a valid API key. Keys are scoped per client, revokable instantly, and never stored in your codebase.

Rate limiting built in

All endpoints are rate limited per API key. Protects against abuse and keeps your costs predictable.

Transparent data use

Customer conversations are processed for scoring and evaluation only. Never used for AI model training.

SOC 2 and GDPR certification in roadmap. We will only claim compliance when it is verified and real.

Get started

Support quality, monitored continuously.

Connect your existing helpdesk via API or outbound webhook. No workflow replacement. No agent retraining. Live in under 30 minutes.

Request access Schedule onboarding →

Works with your existing helpdeskAPI + webhook readyLive in under 30 minutesNo workflow replacement