The Problem We Kept Hearing
"We don't have enough website traffic."
That's what our customers kept telling us. They'd buy Warmly's Inbound Agent, see it convert visitors into meetings, and then hit a wall. Not enough people on their site to work with.
One customer - a Series B SaaS company doing about $3M ARR - told us: "The Inbound Agent is incredible. When someone's on our site, it converts. But we're getting maybe 2,000 unique visitors a month. That's not enough to build pipeline."
Another said: "We'll come back when we have more traffic. Right now, inbound alone isn't going to get us to our number."
We heard some version of this dozens of times. And it kept bugging us, because the underlying logic was wrong. These companies didn't have a traffic problem. They had an awareness problem.
Think about it. If you're a B2B SaaS company selling to mid-market, your total addressable market is probably 10,000 to 30,000 companies. Maybe less. Most of those companies don't know you exist yet. They're not going to magically show up on your website. You need to go find them.
That's why we built the TAM Agent.
Quick Answer: What Is a TAM Agent?
A TAM Agent is an AI system that builds your total addressable market from scratch, scores every account for intent and ICP fit, identifies the buying committee at each company, and activates those contacts across your outbound channels - HubSpot, LinkedIn Ads, and email sequences. Warmly's TAM Agent combines company data from 30M+ businesses, intent signals from 37K+ topics, and a contact database of 220M+ people to find the accounts that should know about you but don't yet. It's the upstream engine that feeds your inbound motion with the right accounts.
The Math: Your TAM Is Finite (and That's a Good Thing)
Here's an exercise we run with every new customer. Work backwards from your revenue goal.
Let's say you need $5M in new ARR this year.
If your average deal is $50K:
- You need 100 new customers
- At a 0.8% account-to-customer conversion rate (which is realistic for B2B SaaS), that's 12,500 accounts in your pipeline funnel
- At a generous 2% of TAM entering your funnel annually, you need a TAM of about 625,000 - no, wait. Let's be real. You need to actively work about 12,500 accounts.
If your average deal is $20K:
- You need 250 new customers
- At the same 0.8% rate, that's 31,250 accounts to work
Here's the point: your TAM is finite. It's 10K to 30K companies. That's small enough to actually work. Small enough to know every account. Small enough to personalize outreach for. Small enough to own.
Most sales teams don't think this way. They're either:
- Spraying cold emails at millions of contacts and hoping something sticks, or
- Waiting for inbound and hoping enough people find their website
Both strategies leave money on the table. The right approach is to map your entire TAM, score every account for fit and intent, and then systematically move them through a journey:
Unaware → Aware → Engaged → Pipeline → Customer
The TAM Agent handles steps one through three. It finds the accounts that should know about you, makes them aware through LinkedIn Ads and outbound sequences, and engages them until they're ready for a conversation.
What the TAM Agent Does: 5 Steps
Here's a walkthrough of how the TAM Agent works, end to end. I recorded a full Loom walkthrough if you want to see it live.
The TAM Agent pulls accounts from multiple sources:
- Your CRM - existing accounts from HubSpot or Salesforce that you want to re-score and enrich
- Website visitors - companies that have already visited your site (de-anonymized by Warmly)
- Domain imports - paste a list of domains you're interested in (competitor customers, event attendee lists, target account lists)
- Third-party signals - companies showing buying intent for topics relevant to your product
You can start with a hundred accounts or a hundred thousand. The agent doesn't care - it'll process and score all of them.
Step 2: Score Intent with ML
This is where most tools fall apart. They give you a black-box "intent score" and say "trust us." We think that's garbage.
Warmly's intent scoring is completely transparent. For every account, you can see exactly why it scored the way it did:
- Session velocity - how many website sessions in the last 7/14/30 days, and is that accelerating?
- Unique visitors - how many distinct people from that company visited?
- Session quality - are they browsing the blog or spending 12 minutes on your pricing page?
- Third-party intent - are they researching topics related to your product on other sites?
- Engagement signals - have they opened emails, clicked ads, engaged on LinkedIn?
Each signal is visible. Each contributes a weighted score. You can see the math. No black boxes, no "proprietary algorithms" you can't inspect.
Why this matters for AI lead scoring: When your SDRs can see why an account is scored high, they trust the data and actually act on it. When it's a black box, they ignore it. We've seen this pattern with every customer who's migrated from 6sense or Demandbase - transparent scoring drives adoption.
Step 3: Qualify with AI Enrichment
Once accounts are scored, the TAM Agent enriches each one with AI-powered qualification:
- Custom fields - define any field you need (e.g., "Does this company sell to enterprise?", "Do they have an outbound sales motion?") and the AI fills it in with reasoning
- ICP Tier classification - our "easy button." The agent classifies every account as Tier 1, Tier 2, or Not ICP based on your ideal customer profile, and shows its reasoning for each classification
This isn't just a yes/no filter. The AI writes a sentence explaining why it made the classification. Something like: "Tier 1 - B2B SaaS, 230 employees, has SDR team of 8, active on G2 comparing sales engagement platforms, recently hired VP of Sales Development." Your reps can read the reasoning and decide whether to override.
This is ICP scoring automation that actually explains itself.
tep 4: Find the Buying Committee
This is the step that changes the game. The TAM Agent doesn't just identify companies - it finds the specific people you need to talk to.
For each account, it:
- Checks your CRM first - if you already have contacts at that company, it uses them
- Searches 220M+ contacts - finds people matching your buying committee personas (Decision Maker, Champion, Influencer, Approver)
- Assigns confidence scores - each contact gets a confidence score for how well they match the persona
- Labels by persona - so your reps know exactly who to reach and what angle to use
The buying committee for a typical mid-market deal might look like:
| Persona | Example Match | Confidence |
|---|
| Decision Maker | VP of Sales, Acme Corp | 94% |
| Champion | Director of SDR, Acme Corp | 91% |
| Influencer | Director of Marketing, Acme Corp | 87% |
| Approver | CEO, Acme Corp | 82% |
You're not blasting a generic email to "info@acme.com." You're reaching the VP of Sales with a message about pipeline generation, the Director of SDR with a message about rep productivity, and the CMO with a message about account-based marketing. Each person gets a relevant angle.
This is buying committee identification software that actually scales. Most teams try to do this manually - a rep spends 15 minutes per account on LinkedIn finding the right people. The TAM Agent does it for thousands of accounts in minutes.
Step 5: Activate Everywhere
The last step is getting these contacts into your outbound channels:
- HubSpot sync - contacts are created or updated in HubSpot with persona labels, ICP tier, intent score, and all enrichment data. Your reps see everything in their CRM without switching tools.
- CSV export for LinkedIn Ads - export a perfectly formatted CSV for LinkedIn Ads matched audiences. When every contact in your audience is a real buyer at an ICP account, your ad spend stops being wasted on random impressions.
- Email sequences - push contacts into Outreach sequences or HubSpot sequences with persona-specific messaging
The TAM Agent doesn't just build a list. It builds the infrastructure for your entire outbound AI agent motion - the right accounts, the right people, the right context, pushed to the right channels.
The Signals That Power It
The TAM Agent doesn't rely on a single data source. It pulls from a wide range of company-level and contact-level signals to build the most complete picture possible.
Company-Level Signals
| Signal Category | Source | Refresh Frequency | What It Tells You |
|---|
| Hiring trends | 30M+ companies tracked | Weekly | Growing teams = growing budget. A company hiring 5 SDRs is about to invest in sales tools. |
| Intent topics | Bombora (37K+ topics) | Daily | What subjects they're researching across the B2B web |
| Company news | SEC filings, press releases | Daily | Fundraising, M&A, leadership changes |
| GitHub activity | Public repositories | Weekly | Tech stack signals, engineering investment |
| Social media | LinkedIn company pages | Weekly | Product launches, culture signals |
| Website intelligence | Warmly pixel | Real-time | Which pages they visit, how often, session quality |
| Product reviews | G2, TrustRadius, Capterra | Weekly | Comparing competitors in your category |
| SEO/traffic estimates | SimilarWeb data | Monthly | Website growth trends, marketing investment |
Contact-Level Signals
| Signal Category | Source | Refresh Frequency | What It Tells You |
|---|
| LinkedIn posts | Public activity | Bi-weekly | What topics they care about (great for personalization) |
| LinkedIn comments | Public activity | Bi-weekly | Who they engage with, what resonates |
| Job changes | LinkedIn profiles | Weekly | New role = new budget, new priorities |
| Podcast appearances | Public directories | Monthly | Thought leadership topics, speaking themes |
| Twitter/X activity | Public posts | Weekly | Real-time opinions and interests |
| YouTube | Public videos | Monthly | Conference talks, product demos |
|
|
The key insight about intent data for outbound sales: No single signal is reliable on its own. Bombora intent alone has a high false positive rate. Hiring data alone doesn't tell you timing. Website visits alone might be a researcher, not a buyer. The TAM Agent combines all of these into a composite score that's far more predictive than any individual signal.
Real Results: The Drift Use Case
Here's a concrete example of what happens when you point the TAM Agent at a specific opportunity.
When Drift got acquired and started sunsetting features, we knew there were hundreds of companies suddenly looking for a replacement. Classic TAM expansion strategy - a competitor exits, and their customers become your TAM.
Here's what we did:
- Imported 169 Drift customer domains into the TAM Agent
- Let it score and classify - filtered down to ICP Tier 1 and Tier 2 accounts
- Found the buying committee at each qualified account - Decision Makers, Champions, Influencers
- Exported to LinkedIn Ads - created a matched audience of real buyers at companies actively looking for a Drift replacement
The result: 11% click-through rate on LinkedIn Ads.
For context, the average LinkedIn Ads CTR is 0.4-0.6%. We hit 11%. That's not a typo.
Why? Because every single impression in that audience was hitting a real buyer - someone with budget authority or influence - at a company that was actively looking for exactly what we sell. No waste. No impressions on random employees. No broad targeting and hoping for the best.
This is what happens when your audience is built from buying signal detection and buying committee mapping instead of loose firmographic targeting.
Full Funnel: TAM Agent + Inbound Agent
The TAM Agent doesn't replace our Inbound Agent. They're two halves of the same system.
TAM Agent = everything pre-site. It handles the outbound AI agent motion - finding accounts, scoring intent, mapping buying committees, running LinkedIn Ads, and sending outbound sequences. Its job is to make the right people aware of you and drive them to your site.
Inbound Agent = on-site conversion. Once those people land on your site, the Inbound Agent takes over - AI chat, retargeting, email nurture, and real-time engagement. It already knows who they are (because the TAM Agent mapped them), so it can personalize instantly.
The Brain connects everything. It's the shared intelligence layer - a context graph that remembers every interaction, every signal, every touchpoint. When someone from a TAM Agent audience clicks a LinkedIn Ad and lands on your pricing page, the Brain knows their ICP tier, their buying committee role, their intent score, and their engagement history. The Inbound Agent uses all of that context to have a relevant conversation.
This is what full-funnel account-based marketing AI actually looks like. Not a small slice of the funnel with one tool for ads and another for email and another for chat. Full context, from first awareness to closed deal.
How Is This Different from ZoomInfo, 6sense, or Demandbase?
I'll be direct. Here are the real differences - not marketing speak.
vs. ZoomInfo: ZoomInfo is a contact database. A really good one. But it doesn't score intent transparently, doesn't classify ICP with AI reasoning, and doesn't build buying committees automatically. You get a list of people and you're on your own to figure out who matters and when to reach out. The TAM Agent does the thinking for you.
vs. 6sense: 6sense has strong intent data and predictive scoring, but it's a black box. You can't see why an account scored the way it did. Their buying committee features require manual setup. And their pricing starts at $55K+/year with complex implementation timelines. The TAM Agent is transparent, automated, and available at a fraction of the cost.
vs. Demandbase: Similar to 6sense - enterprise-focused ABM platform with strong ad targeting but opaque scoring, complex setup, and enterprise pricing. The TAM Agent gives you the same capability (intent scoring, buying committee, ad activation) without the 6-month implementation.
The real difference: These tools were built for a world where you have dedicated ops teams to configure, maintain, and interpret them. The TAM Agent was built for teams that want to press a button and get results. Import accounts, let the agent score, qualify, find people, and activate. That's it.
What's Coming Next
We're actively building:
- Native LinkedIn Ads integration - one-click audience sync directly from the TAM Agent to LinkedIn Campaign Manager. No more CSV exports.
- Native Meta Ads integration - same one-click sync for Meta/Facebook Ads audiences
- More third-party signal sources - we're adding new company and contact signal providers to make intent scoring even more accurate
- Automated activation loops - the TAM Agent will automatically refresh audiences and sequences as intent scores change, keeping your outbound always current
Try It
The TAM Agent is available now for all Warmly customers.
Book a demo to see it in action on your actual TAM.
Watch the full walkthrough to see how it works step by step.
If you're already a Warmly customer, reach out to your account manager - they can get you set up in a single session.
Frequently Asked Questions
What is a TAM agent?
A TAM agent (Total Addressable Market agent) is an AI-powered system that builds, scores, and activates your total addressable market automatically. Instead of manually researching companies and contacts, a TAM agent identifies every company that fits your ideal customer profile, scores them for buying intent, finds the right people to contact, and pushes them into your outbound channels like HubSpot, LinkedIn Ads, and email sequences.
How does Warmly's intent scoring work?
Warmly uses a transparent, multi-signal intent scoring model that combines website session velocity, unique visitor counts, session quality metrics, third-party Bombora intent data, and engagement signals like email opens and ad clicks. Every signal is visible - you can see exactly which factors contributed to each account's score and how much weight each carries. This is fundamentally different from black-box scoring used by tools like 6sense and Demandbase, where you can't inspect the reasoning.
What is a buying committee and how does the TAM Agent find one?
A buying committee is the group of people at a company who influence or decide a purchase - typically a Decision Maker (VP/C-level with budget), a Champion (the person pushing for the tool internally), an Influencer (someone who shapes evaluation criteria), and an Approver (often CEO at smaller companies). The TAM Agent finds buying committees by first checking your CRM for existing contacts, then searching a database of 220M+ contacts to match people by title, seniority, and department to each persona, assigning confidence scores for each match.
How many contacts does Warmly have access to?
Warmly's contact database includes over 220 million professional contacts with verified email addresses, job titles, company affiliations, and LinkedIn profiles. The database is continuously refreshed with new contacts added weekly and existing records verified against multiple data providers using a consensus-based approach.
Can I connect the TAM Agent to HubSpot or Salesforce?
Yes. The TAM Agent integrates directly with HubSpot and Salesforce. Contacts are synced with full enrichment data including persona labels, ICP tier classification, intent scores, and AI-generated qualification notes. Your reps see everything directly in the CRM without switching between tools.
What signals does the TAM Agent use to score accounts?
The TAM Agent uses company-level signals (hiring trends across 30M+ companies, Bombora intent data for 37K+ topics, company news, SEC filings, GitHub activity, product reviews on G2/TrustRadius, SEO traffic trends, and website visitor behavior) plus contact-level signals (LinkedIn posts and comments, job changes, podcast appearances, and Twitter/X activity). These signals are combined into a composite intent score that's significantly more predictive than any single signal source.
How is the TAM Agent different from ZoomInfo or 6sense?
ZoomInfo is primarily a contact database - it gives you people to call but doesn't score intent transparently or build buying committees automatically. 6sense offers strong intent data but uses opaque, black-box scoring and starts at $55K+/year. The TAM Agent combines transparent intent scoring, automated ICP classification with AI reasoning, buying committee identification with confidence scores, and multi-channel activation — at a fraction of the cost and without the 6-month implementation timeline.
What does ICP tier classification mean?
ICP (Ideal Customer Profile) tier classification is the TAM Agent's AI-powered system for grading how well each account matches your ideal customer. Tier 1 accounts are a strong match across all criteria (industry, company size, sales team structure, tech stack). Tier 2 accounts match most criteria but may have one gap. Not ICP accounts don't fit your profile. The AI provides written reasoning for each classification so your team can verify and override if needed.
Can I use the TAM Agent for LinkedIn Ads?
Absolutely. The TAM Agent exports perfectly formatted CSV files for LinkedIn Ads matched audiences. Because the audience is built from buying committee contacts at ICP-qualified, intent-scored accounts, every impression hits a real buyer - which is why customers see dramatically higher CTRs (one campaign hit 11% CTR versus the 0.4-0.6% LinkedIn average).
What's the difference between the TAM Agent and the Inbound Agent?
The TAM Agent handles everything pre-site - building your target account list, scoring intent, finding buying committees, and running outbound across LinkedIn Ads and email sequences. The Inbound Agent handles on-site conversion - AI chat, retargeting, email nurture, and real-time engagement when visitors land on your website. Together, they cover the full funnel from first awareness to closed deal, connected by The Brain which maintains context across every interaction.
How do I import accounts into the TAM Agent?
You can import accounts four ways: (1) sync directly from your CRM (HubSpot or Salesforce), (2) upload a CSV of company domains, (3) pull from your Warmly website visitor data, or (4) import from third-party signal sources. Most customers start by importing their existing CRM accounts for re-scoring, then add target account lists and competitor customer domains.
How often is the data refreshed?
Signal refresh frequencies vary by type: website visitor data is real-time, Bombora intent data refreshes daily, hiring trends and job change data update weekly, LinkedIn activity scans bi-weekly, and broader market signals like SEO traffic and company news refresh weekly to monthly. Intent scores are recalculated as new signals arrive, so your account prioritization is always current.
Further Reading
TAM Agent Resources
Related Posts