How to Build an AI SDR Workflow: Clay + Instantly + HubSpot in One Afternoon
What if your next SDR could research leads, enrich contact data, personalize emails, launch campaigns, and update the CRM before your first afternoon meeting? That is the idea behind an AI SDR workflow. You do not need to build a fully autonomous sales agent to automate a meaningful part of outbound. With Clay for research and enrichment, Instantly for sending, and HubSpot for CRM control, you can create a practical system that turns raw prospects into qualified outbound opportunities. This tutorial shows how to build that workflow in one afternoon, including the data structure, automation logic, personalization process, and deliverability safeguards.
Tools You’ll Need
1. Clay
Use Clay as the research and enrichment layer.
You can use it to:
- Import company or contact lists
- Enrich company information
- Find missing contact data
- Research websites and public signals
- Generate custom AI research fields
- Create personalized messaging inputs
Clay's current plans include multi-provider waterfalls and Claygent, its AI agent for web research and custom data points.
2. Instantly
Use Instantly as the outbound sending layer.
It handles:
- Cold email campaigns
- Sending accounts
- Campaign sequences
- Reply tracking
- Deliverability workflows
- Integrations and automation
Instantly's integration ecosystem includes Clay, HubSpot, Zapier, webhooks, and API options.
👉 Suggested Read: Instantly.ai Review: Deliverability, Inbox Rotation & Real Reply Rates
3. HubSpot
Use HubSpot as the system of record.
It should contain:
- Lead status
- Contact details
- Company details
- Campaign source
- Reply status
- Qualification status
- Sales owner
- Next action
HubSpot supports HubSpot sales automation through sequences and workflows that can automate follow-up, lead routing, tasks, and other repetitive sales actions.
4. Zapier Optional
Zapier is useful when you need a simple no-code bridge between tools.
For example, you can connect HubSpot and Clay to create records, synchronize contact information, or trigger actions when a CRM event occurs.
Step 1: Define Your Ideal Customer Profile
Do not start by building automation.
Start by defining the type of account you actually want.
Create a simple ICP specification:
|
Field |
Example |
|
Industry |
B2B SaaS |
|
Company Size |
20–200 employees |
|
Geography |
United States, UK, Canada |
|
Buyer |
VP Sales, Head of Growth |
|
Technology |
HubSpot, Salesforce |
|
Trigger |
Hiring SDRs or launching outbound |
The more specific your ICP is, the better your AI SDR workflow will perform.
A weak ICP produces a large list of irrelevant contacts.
A strong ICP gives your enrichment and personalization system useful context.
Step 2: Create Your Clay Table
The Clay table is the central workspace for your lead enrichment workflow.
Create columns for:
- Company name
- Company domain
- Contact name
- Job title
- Work email
- LinkedIn URL
- Employee count
- Industry
- Technology used
- Recent trigger
- ICP score
- Research summary
- Personalization angle
- Email draft
- Campaign status
A simple Clay table might look like this:
Company → Domain → Contact → Role → Enrichment → AI Research → Personalization → Email
Clay Tables Tutorial: The Basic Workflow
- Import your target companies or contacts.
- Add enrichment providers.
- Add company and contact data.
- Run AI research on qualified records.
- Generate a personalization angle.
- Create an email draft.
- Review the output.
- Send approved records to Instantly.
Clay's multi-provider waterfall capability allows teams to use different providers sequentially when the first source does not return usable data. That can be useful when your target market is difficult to enrich or your ICP requires more than basic contact information. (Clay waterfall enrichment)
Step 3: Add AI Research
This is where your workflow becomes more than a simple data scraper.
Instead of asking AI to write a generic email, give it a specific research task.
For example:
Research this company and identify one recent business signal that could create a need for outbound sales automation.
The output could identify:
- New funding
- Rapid hiring
- New market expansion
- New product launch
- Sales team growth
- Website messaging changes
- New technology adoption
Then create a structured field called:
Outbound Trigger
For example:
“The company is hiring five SDRs, suggesting the sales team is expanding outbound capacity.”
That is much more useful than:
“I noticed your company is growing.”
Step 4: Create AI Personalization at Scale
Your AI personalization prompt should use real data.
A useful structure is:
Input:
- Company description
- Contact role
- Recent trigger
- Technology used
- Research summary
Instruction:
Write one concise opening sentence that connects the prospect's specific business situation to the problem our product solves. Do not use generic compliments. Do not mention information that is not included in the research data.
This creates a more controlled AI cold email workflow.
Example
Research signal:
Company recently hired three SDRs.
Weak personalization:
“I saw that your company is growing rapidly.”
Better personalization:
“Noticed your SDR team is expanding—many teams hit a data and personalization bottleneck when outbound volume increases.”
The second version is more specific and creates a logical reason for the outreach.
Step 5: Add an ICP Score
Do not send every enriched record to Instantly.
Create a simple scoring system.
For example:
- ICP match: 0–40 points
- Seniority: 0–20 points
- Trigger signal: 0–20 points
- Data quality: 0–20 points
Then define:
- 80–100: Send to outbound
- 60–79: Manual review
- Below 60: Do not send
This is one of the most important steps in an AI SDR workflow.
Automation should improve prioritization, not simply increase the number of emails sent.
Step 6: Send Qualified Leads to Instantly
Once your Clay table contains qualified prospects, send only the approved records to Instantly.
Your transfer should include:
- First name
- Last name
- Company
- Job title
- Personalization line
- Research trigger
- Campaign name
There are several ways to connect Clay and Instantly, including native integrations, webhooks, API-based workflows, or automation platforms. Instantly's integration documentation includes dedicated resources for Clay, HubSpot, webhooks, and Zapier connections.
Example Data Flow
Clay:
John → VP Sales → SaaS company → Hiring SDRs → Personalization line
↓
Instantly:
Campaign: SaaS Outbound Q3
↓
Email:
Personalized opening + relevant problem + clear CTA
Step 7: Build the Cold Email Campaign
Keep the campaign simple.
A basic sequence might include:
Email 1 — Relevant Observation
Use the research signal to explain why you are reaching out.
Email 2 — Problem Expansion
Follow up with a specific operational problem related to the first message.
Email 3 — Useful Insight
Share a practical observation, benchmark, or relevant resource.
Email 4 — Breakup or Close-the-Loop Email
Give the prospect a simple way to respond.
Do not make every email sound like a different AI model wrote it.
Your campaign should have a consistent voice.
Step 8: Set Up Cold Email Deliverability
Your AI SDR workflow is useless if your emails do not reach the inbox.
Before sending at scale:
- Use properly configured sending domains
- Configure SPF, DKIM, and DMARC
- Warm up new sending infrastructure carefully
- Avoid sudden volume spikes
- Validate email addresses
- Monitor bounce rates
- Keep email copy relevant
- Provide a clear opt-out path where appropriate
Do not treat deliverability as a feature you can automate away.
Your sending infrastructure, domain reputation, list quality, and campaign behavior all matter.
Step 9: Send Reply Signals to HubSpot
The next step is connecting outbound activity to your CRM.
When a prospect replies, update HubSpot with:
- Contact status
- Reply classification
- Campaign name
- Last activity
- Sales owner
- Follow-up task
A simple status structure might be:
New → Contacted → Replied → Qualified → Meeting Booked → Opportunity
HubSpot workflows automation can automate tasks, routing, notifications, and data updates across the customer journey.
Step 10: Create the HubSpot Outbound Sequence
Use HubSpot for the human follow-up after a positive response.
For example:
Trigger: Prospect replies positively.
Action 1: Update lifecycle or lead status.
Action 2: Assign the lead to a sales owner.
Action 3: Create a follow-up task.
Action 4: Send an internal notification.
Action 5: Add the prospect to the appropriate sales sequence.
HubSpot sequences are designed for timed sales follow-up, while workflows can automate broader CRM processes such as routing, tasks, and data updates. (HubSpot sales automation)
The Complete AI SDR Workflow
Here is the final process:
1. Define ICP
Identify the companies and roles you want to reach.
2. Build the Clay Table
Import accounts and create enrichment fields.
3. Enrich Leads
Find company, contact, technology, and firmographic data.
4. Research Accounts
Use AI to identify relevant business signals.
5. Score Prospects
Prioritize accounts based on fit and intent.
6. Generate Personalization
Create a specific, research-based opening line.
7. Send to Instantly
Launch the approved campaign.
8. Monitor Replies
Track positive, negative, and out-of-office responses.
9. Sync to HubSpot
Update CRM records and sales statuses.
10. Trigger Human Follow-Up
Let sales reps handle qualified conversations.
Common Mistakes to Avoid
Automating Before Defining the ICP
More automation does not fix poor targeting.
Sending AI-Generated Emails Without Review
AI can produce inaccurate or awkward personalization. Review samples before launching a campaign.
Sending Every Lead to the Outreach Tool
Use scoring and filtering before sending prospects to Instantly.
Ignoring Data Duplicates
Duplicate records can waste enrichment credits and create confusing CRM records.
Treating AI as a Replacement for Sales Judgment
The best workflow automates research and repetitive tasks while keeping humans involved in qualification and relationship-building.
Expert Tip: Build a Human-in-the-Loop Workflow
The best AI SDR workflow is not fully autonomous.
Use AI for:
- Research
- Data enrichment
- Lead scoring
- Drafting
- Classification
- Routing
Use humans for:
- Final campaign approval
- High-value account review
- Positive reply handling
- Qualification
- Sales conversations
This creates a more reliable system than allowing AI to send unlimited messages without supervision.
Conclusion
A useful AI SDR workflow does not require a complicated autonomous sales agent. With Clay handling research and enrichment, Instantly managing outbound campaigns, and HubSpot tracking CRM activity and follow-up, a founder or RevOps team can build a practical sales automation system in a single afternoon. The key is to automate the repetitive work without removing human judgment from the process. Start with a narrow ICP, enrich only qualified prospects, use AI to create relevant context, and send positive replies directly into a human-led sales process.
Frequently Asked Questions
What is an AI SDR workflow?
An AI SDR workflow is a sales automation process that uses artificial intelligence to research prospects, enrich data, prioritize leads, personalize outreach, automate follow-ups, and route qualified responses to sales teams.
How do you automate outbound sales?
You can automate outbound by connecting lead sourcing, data enrichment, AI research, personalization, email sending, reply tracking, and CRM updates into a single workflow.
Can Clay connect to Instantly?
Yes. Clay and Instantly can be connected through available integrations and automation methods such as native connections, webhooks, APIs, and automation platforms.
Should HubSpot or Instantly send cold emails?
For a dedicated cold outbound workflow, Instantly can handle campaign sending while HubSpot acts as the CRM and sales system of record. The right setup depends on your compliance requirements, sales process, and existing technology stack.
Is an AI SDR workflow fully automated?
It can be, but a human-in-the-loop model is usually safer and more effective. AI can handle research, enrichment, and drafting, while humans review important campaigns and manage qualified conversations.