← Back to blog
Build idea SalesClayOllamaMuse Glimmer

Score Inbound Leads in Clay with Muse Glimmer

Build idea, not a verified tutorial. This was drafted by an automation Daniel built. Tool interfaces change often — check the current docs before relying on the steps.

Reps waste time on low-fit leads because every new submission looks the same in a spreadsheet. You need something to read each lead's context and rank them before anyone picks up the phone. Meta just released Muse Glimmer — a free, 30-billion-parameter AI model you can run on your own laptop — so you can do that scoring without paying per API call.

TRIGGERORCHESTRATEAI / ENRICHOUTPUTNew lead via webhookClay tableWaterfall enrichmentMuse Glimmer scoresfitTier tag + rep alert
How it flows
1

Install Ollama and pull Muse Glimmer

Ollama is a free desktop app that lets you run AI models on your own computer — no cloud account needed. Install it from ollama.com, then open your computer's terminal (the text-input app on Mac or Windows) and run the two commands below. The first downloads Muse Glimmer (about 18 GB, so give it time). The second starts it as a local server your other tools can talk to over the internet address `localhost:11434`.

Terminal · downloading and serving the model
1 Step 1 — download the model
ollama pull muse-glimmerCopy
2 Step 2 — confirm it's running
ollama serveCopy
3 Leave this terminal window open
Ollama must keep running in the background the whole time Clay calls it. You can minimise the window — just don't close it. Muse Glimmer on Ollama →
2

Create a Clay table with a webhook source

Clay is a spreadsheet-like tool where each row is a lead and each column can call an API or run AI automatically. A webhook is a special URL — when your form or CRM sends lead data to that URL, Clay creates a new row instantly. Sign up at clay.com (free tier available), then follow the steps below.

Clay · creating the webhook-triggered table
1 Start a new table
+ Create New
2 Choose how data arrives
IC Import CSVSFP Search for peopleSFC Search for companiesIDF Import data from Webhook
3 Clay generates a unique URL — copy it
https://api.clay.com/v1/sources/webhook/YOUR_TABLE_IDCopy
4 Paste it into your lead form
In your form tool (Typeform, HubSpot Forms, or any other), find the webhook or notification settings and paste this URL. Every new submission will now create a row in Clay. Clay webhook guide →
3

Add waterfall enrichment columns

Before scoring, Clay needs real data about each lead — company size, industry, website. A waterfall means Clay tries Provider A first; if that misses, it tries Provider B automatically. You don't have to manage this — you just pick the providers once. Click the column-add button to get started.

Clay · adding enrichment columns
1 Add a new enrichment column
Add enrichment
2 Search for company data
Company Enrichment
EC Enrich CompanyClay built-in
FC Find CompanyApollo
CL Company LookupClearbit
3 Set up the waterfall for company size
Clay built-in→Apollo→Clearbit
Only run if company_size is empty
4 Repeat for these fields
IndustryClay built-in → Apollo
WebsiteClay built-in
Employee countClay built-in → Clearbit
Run on new row
4

Add a Claygent column to read the lead's website

Claygent is Clay's built-in AI research agent. Give it a plain-English question and it visits the lead's website and returns a structured answer. This gives Muse Glimmer richer context to score against. Click Add enrichment, search for Claygent, then write a prompt.

Clay · Claygent setup
1 Open the enrichment menu again
Add enrichment
2 Find Claygent
Claygent
C ClaygentClay AI agent
3 Write your research question (plain English)
Visit {{website}}. In 2 sentences, describe what this company sells and who their main customer is.
Returns"Acme sells project-management software to construction firms. Their main buyer is the general contractor."
4 Confirm and run
Save and run all rows
5

Call Muse Glimmer to score each lead

Now you connect Clay to your local Muse Glimmer. Clay's HTTP API enrichment column lets any column call an external URL and store the response. Your Ollama server is listening at `http://localhost:11434` — but Clay (running in the cloud) can't reach your laptop's localhost directly. Use a free tunnel tool like ngrok (`ngrok http 11434`) to get a public URL first, then paste that into Clay. The moment a new row arrives, Clay sends the lead's data to Muse Glimmer and writes back a score.

Clay · HTTP API enrichment column → Muse Glimmer
1 Add one more enrichment column
Add enrichment
2 Search for HTTP API
HTTP API
HA HTTP APIClay built-in
3 Configure the HTTP call
MethodPOST
URLhttps://YOUR-NGROK-URL.ngrok-free.app/api/chat
Header: Content-Typeapplication/json
Run on new row
4 Paste this as the JSON body
{ "model": "muse-glimmer", "messages": [{ "role": "user", "content": "Score this lead 1–10 for fit with a B2B SaaS tool aimed at ops teams. Return JSON only: {score, tier, reason}. Lead data: company={{company}}, size={{employee_count}}, industry={{industry}}, what they do: {{claygent_summary}}" }] }
Returns{ "score": 8, "tier": "Tier 1", "reason": "Ops-heavy construction firm, 200+ staff, strong ICP match." }
6

Parse the score and add a routing action

Muse Glimmer returns raw JSON text. Add a formula column (type `=` in any blank column header, then use the formula editor) to extract the `tier` field from that text. Then add one final enrichment column — Send Email or Slack message — so your team gets an alert only when tier equals "Tier 1". Everything else sits quietly in Clay for weekly review.

Clay · formula + routing columns
1 Formula column — extract the tier
Formula=JSON_EXTRACT(muse_glimmer_response, "tier")
OutputTier 1 / Tier 2 / Tier 3
Run automatically
2 Add a conditional action column
Add enrichment
3 Search for Slack
Slack message
SSM Send Slack MessageSlack
4 Only fire when it's worth it
Conditiontier equals Tier 1
Message🔥 New Tier 1 lead: {{company}} — Score {{score}}. {{reason}}
Skip if condition is false
5 What a finished row looks like
Acme Construction Co.Tier 1
Score8 / 10
TierTier 1
ReasonOps-heavy firm, 200+ staff, strong ICP match
Email✓ verified

What you get

You end up with a Clay table where every new inbound lead is automatically enriched, researched, and scored by Muse Glimmer running free on your own machine. Tier 1 leads fire a Slack alert the moment they arrive; everything else waits quietly for review. Rough effort: ~2–3 hours to set up. Free: Clay free tier, Ollama free, Muse Glimmer Apache 2.0. Paid only if you exceed Clay's free enrichment credits or need a paid ngrok plan for a persistent public URL..

Sources