AI Sales Reps

Think of an AI Sales Rep as a new hire joining your team -- except this one learns in minutes, not months. Before they can represent your company, they need to understand your business from every angle: what you sell, who you sell to, how you talk to prospects, what makes you different, and what they should never say. The setup process is designed to feel like you're actually onboarding a real person -- sitting across from them, answering their questions, watching them connect the dots in real time.

The better trained your sales rep is, the better it performs. And the training process itself is what makes this possible.

How Sales Reps Work

Your AI Sales Rep handles two core jobs:

• Writing personalized first emails for campaigns -- each one tailored to the specific person you're reaching out to
• Replying to responses -- when someone writes back, your sales rep continues the conversation naturally, just like a real person would

Every email your sales rep writes draws on everything it knows about your company. That's why the setup process matters so much -- the more your sales rep understands about your business, the more authentic and effective its emails become.

Setting Up a New Sales Rep

Guided Setup (Recommended)

The best way to create a new sales rep is through the guided setup. This is the heart of the product -- a 6-step process that walks you through training your rep the way you'd onboard your best new hire.

1. Go to Agents in your dashboard
2. Click "Create with AI"
3. Enter your company's website URL

From there, the system reads your website and builds an initial understanding of your business. Then the real training begins.

Step 1: Website Analysis

Enter your website URL, and the AI will read your site to learn about your company -- what you sell, who your customers are, and what makes you different. You can also add any additional context that might not be on your website.

This is like handing a new hire your website on their first day and saying "Start here." The AI scrapes and analyzes your site, pulling out everything it can about your company, products, positioning, and audience.

Step 2: Review What We Learned

You'll see a summary of what the AI extracted from your site -- your company overview, products, target audience, competitive advantages, and pricing. Everything is editable. If something is off, just correct it.

Think of this as your new hire coming back to you and saying, "Here's what I think I understand about the company." You can correct anything, add details they missed, and make sure the foundation is solid before moving on.

Step 3: The Interview -- Where It Gets Dynamic

This is where training your sales rep goes from filling out forms to having an actual conversation.

Once the AI has a baseline understanding of your business, it reviews what it knows and identifies the gaps -- things it still isn't confident about, areas where the website was vague, or critical sales knowledge that no website ever covers. Then it asks you about them, one question at a time.

How the conversation works:

The AI generates 5-8 targeted questions based on what's actually missing from your specific situation. These aren't generic questions from a template. If your website had strong pricing information but said nothing about how you handle objections, the questions will focus on objections, not pricing. If your competitive positioning was clear but your tone was ambiguous, it'll ask about tone.

Each question falls into a category -- things like:

• Tone & Communication Style -- "How would you describe your brand voice?"
• Objection Handling -- "What are the top objections you hear from prospects?"
• Competitive Landscape -- "Who do people usually compare you to?"
• Call-to-Action Strategy -- "What's the one thing you want prospects to do after reading your email?"
• Pricing Communication -- "How openly do you want your rep to discuss pricing?"
• Forbidden Topics -- "Are there any topics or claims your rep should never bring up?"

Follow-up questions that go deeper:

Here's what makes this feel like a real conversation rather than a survey: when your answer reveals something interesting or opens up a new area, the AI generates a follow-up question on the spot and inserts it right into the conversation.

For example, if you mention that prospects often compare you to a specific competitor, the AI might immediately follow up with: "What's the main reason customers choose you over [that competitor]?" Or if you say your tone should be casual, it might ask: "How casual? Like a friend texting, or more like a friendly colleague in a Slack message?"

This back-and-forth is what makes the training so effective. You're not just filling in blanks -- you're having an intelligent conversation where the AI is actively listening, adapting, and drilling deeper based on what you tell it.

What happens behind the scenes:

Every answer you give is immediately synthesized into concrete configuration for your sales rep. When you describe your tone, the agent's core instructions update. When you mention an objection, a new objection-handling rule is created with an approved response. When you say something should never be mentioned, it gets added to the guardrails. You never have to worry about which settings to change -- you just talk, and the AI translates your words into the right settings behind the scenes.

Your controls during the interview:

• Quick answers -- each question comes with suggested answers you can select with one click, or you can type your own response
• "Why are we asking this?" -- click to see exactly how your answer will improve the sales rep's performance
• Skip -- skip any question you're not sure about
• "Done, I've shared enough" -- end the interview early if you feel the rep has what it needs

The "What I Know So Far" Panel -- Watch Your Rep Learn in Real Time

Throughout the entire setup process, there's a live knowledge panel on the right side of the screen that shows what your sales rep understands right now, written in first person from the rep's perspective.

This panel is organized into 8 knowledge areas:

• Company Overview -- who you are and what you do
• Products & Services -- what you sell and how it works
• Value Proposition -- why you're better than the alternatives
• Target Audience -- who your ideal customers are
• Sales Approach -- how your rep should communicate and close
• Objection Handling -- how to respond to common pushback
• Guardrails -- what the rep should never say or do
• Training Materials -- reference documents and email samples

Every time you answer a question in the interview, the relevant sections of this panel update in real time -- you can literally watch the text stream in as your rep processes what you just told it. If you answered a question about your competitive advantages, the Value Proposition section updates. If you described how to handle a pricing objection, the Objection Handling section fills in.

It's like watching a new hire take notes during training, except these notes are instantly operational -- they directly shape how every email gets written.

Correct anything on the fly: At the bottom of the knowledge panel, there's a text input where you can type corrections at any point, like "Actually, our price is $299, not $399" or "Don't forget we also offer a free tier." The relevant sections regenerate immediately to incorporate your correction.

Step 4: Set Your Sales Playbook

Based on everything from your interview and website analysis, the system pre-fills your sales rep's playbook:

• Common objections and approved responses -- so your sales rep handles pushback consistently, using the exact language you want
• Things to never say -- topics, phrases, or claims your sales rep should avoid
• Call-to-action -- what your sales rep should encourage prospects to do (book a demo, start a trial, etc.)
• Email length -- how long or short you want the emails to be

These are pre-filled based on your earlier answers, but you can always edit, add, or remove any of them. If you mentioned during the interview that prospects always ask about uptime guarantees, you'll find that objection already listed here with an approved response ready to go.

Step 5: Add Training Materials

Upload documents that help your sales rep go deeper -- sales playbooks, product sheets, FAQs, case studies. These are exactly the kind of materials you'd hand a new hire on their first day. You can also connect your Gmail and select real emails that represent your ideal writing style. Your sales rep uses these as reference material when crafting emails.

This step is optional. You can always come back and add more materials later.

Step 6: Review and Launch

See a full summary of everything your sales rep has been configured with, give it a name, and launch it. You can click back to any previous step to make changes before creating.

Quick Create

If you already know exactly how you want your sales rep configured, you can use Quick Create to set up a blank one and fill in the details manually. Click the dropdown arrow next to "Create with AI" and select "Quick Create."

What Your Sales Rep Knows

Once your sales rep is set up, you can see a comprehensive summary of everything it knows about your company by opening the "What I Know" panel. Click the knowledge panel button in the header to toggle it open on the right side of the screen.

This summary is written from the sales rep's perspective -- like a real hire explaining back to you what they've learned during onboarding. It pulls together information from every source:

• Your business context (what you sell, who it's for, why you're better)
• Communication style and instructions
• Objection handling strategies
• Guardrails and things to never say
• Uploaded documents and email samples
• Answers from your setup interview

Generating the Summary

If your sales rep doesn't have a summary yet, click "Generate Knowledge Summary" to create one. The AI will review all of the data and write a clear, organized summary.

Updating with Natural Language

Here's the most powerful part: you can update your sales rep's knowledge just by telling it what to change, in plain English.

At the bottom of the summary, there's a text input where you can type things like:

• "We just launched a new enterprise plan at $499/month"
• "Never mention competitor X by name"
• "Our main differentiator is our 24/7 support team, not price"
• "We no longer offer monthly billing, only annual"

When you send an update, two things happen:

1. The written summary updates to reflect the new information
2. The relevant settings update automatically -- if you mention a new pricing tier, the pricing field updates; if you add a new "never say" rule, it gets added to the guardrails

This means you don't have to dig through individual tabs and fields to make changes. Just tell your sales rep what's new, and it figures out where to put it. It's the same principle as the interview -- you talk naturally, and the system translates your words into configuration.

Configuration Tabs

Beyond the knowledge summary, you can fine-tune every aspect of your sales rep through the configuration tabs:

Instructions

This is where you control how your sales rep writes emails. It has three main areas:

• Outreach Email Instructions -- how the agent writes personalized cold emails. This is the foundational instruction that shapes every initial outreach email.
• Follow-Up Sequence Instructions -- per-step instructions for follow-up emails. You can configure a different approach for each follow-up step (e.g., step 1 = social proof, step 2 = urgency, step 3 = final value-add).
• Reply Instructions -- how the agent handles inbound replies. This shapes the tone and approach when prospects respond to your emails.
• Maximum email length -- controls how long or short outreach and reply emails should be

Each of these fields supports AI drafting -- click the draft button to have the AI generate initial content based on your business context. Each field can also receive AI-powered suggestions from market intelligence (see below).

Business Context

Detailed information about your company, products, and services:

• What you sell -- your products or services
• Who it's for -- your target audience
• Problems you solve -- the pain points you address
• Why you're better -- your competitive advantages
• Pricing -- your pricing structure (if you want the sales rep to reference it)

You can also link specific products from your company's product catalog to give your sales rep deeper product knowledge.

Guidelines

The rules your sales rep follows:

• Objections and approved responses -- pre-approved answers to common pushback, so your sales rep stays on-message
• Primary call-to-action -- what your sales rep should encourage prospects to do (book a demo, start a trial, etc.)
• Things to never say -- phrases, topics, or claims that are off-limits

Training

Additional materials that make your sales rep smarter:

• Documents -- upload PDFs, Word docs, or text files (sales playbooks, product sheets, FAQs)
• Email samples -- select real emails from your Gmail that represent your ideal tone and style
• Websites -- add website URLs as knowledge sources for the agent to reference

The more reference material you provide, the more context your sales rep has when writing emails.

Market Intelligence

Each sales rep has a Market Intelligence section that shows agent-scoped insights -- intelligence gathered specifically from the campaigns this agent has been running. You'll find this on the agent's page.

The Market Intelligence panel shows:

• Executive summary -- a brief synthesis of how prospects are responding to this agent's outreach
• Average interest level -- the overall engagement score from replies and calls
• Sentiment breakdown -- how replies are distributed across positive, neutral, mixed, and negative

Expandable sections provide deeper detail:

• What resonates -- messaging patterns that generate positive engagement for this agent
• What falls flat -- patterns that trigger negative responses
• Objection themes -- the most common objections prospects raise
• Question themes -- what prospects are asking about
• Market signals -- non-obvious patterns and trends
• Call intelligence -- buying signals, competitive mentions, and call-specific insights from analyzed call transcripts
• Email vs. call divergences -- where what prospects say in email differs from what they say on calls

This section links to the full Insights page where you can explore the data in more depth.

AI-Powered Field Suggestions

As your campaigns run and the system collects reply and call data, market intelligence is synthesized and turned into concrete, field-level suggestions for improving your sales rep's configuration.

Which Fields Receive Suggestions

Suggestions can be generated for five key areas:

1. Outreach email instructions -- how your agent writes cold emails
2. Reply instructions -- how your agent handles inbound replies
3. Follow-up sequence instructions -- per-step follow-up strategy
4. Objection responses -- pre-approved responses to prospect pushback
5. Never-say rules -- forbidden phrases and topics

What a Suggestion Contains

Each suggestion includes:

• Reasoning -- a clear explanation of why this change is recommended
• Intelligence highlights -- the 2-4 specific data points from market intelligence that triggered the suggestion (e.g., "Budget concerns are the #1 objection at 14 replies" or "ROI-first messaging shows high confidence resonance")
• Impact level -- rated as high, medium, or low based on how much improvement is expected
• Before/after diff -- a word-level diff showing exactly what would change. For text fields, you see inserted and removed words highlighted. For array fields (like never-say rules), you see which items would be added or removed. For objection responses, you see new, modified, and unchanged entries.

Accepting or Dismissing

Suggestions appear inline on the configuration page, attached to the field they relate to. For each suggestion, you can:

• Accept -- applies the suggestion immediately. The field updates, and a version snapshot is saved.
• Dismiss -- rejects the suggestion. It's marked as dismissed and won't appear again.

When new suggestions are generated, any previous pending suggestions for the same field are automatically expired.

How Suggestions Are Generated

Suggestions are powered by the Insights system. The process:

1. Email replies and call transcripts are analyzed and aggregated at the agent scope
2. When enough new data accumulates, an agentic AI synthesizes qualitative intelligence
3. That intelligence is compared against your agent's current configuration
4. Specific field-level improvements are proposed based on what the data suggests

The system looks for agent-scoped intelligence first. If there isn't enough data at the agent level, it falls back to product-scoped intelligence, then company-scoped.

Version History & Rollback

Every change to your sales rep's configuration fields is tracked with full version history. This includes changes from:

• Manual edits you make directly
• Accepted AI suggestions
• AI-drafted content
• Reverts to previous versions

You can view the version history for any field and revert to any previous version with one click. This gives you the confidence to accept AI suggestions knowing you can always undo the change if it doesn't work out.

Using Sales Reps in Campaigns

Default Sales Rep

You can set one sales rep as the default. This one is automatically used for any campaign that doesn't have a specific sales rep assigned. To set a default, click the three-dot menu next to a sales rep in the sidebar and select "Set as Default."

Assigning a Sales Rep to a Campaign

When you create a campaign, you'll choose which sales rep handles the emails. Different campaigns can use different sales reps -- for example, you might have one for cold outreach and another for following up with existing leads.

What the Sales Rep Does in a Campaign

Once assigned to a campaign, your sales rep:

1. Writes the first email for each recipient, personalized based on their name, title, company, and any other information available
2. Handles replies -- when someone responds, your sales rep reads the full conversation thread and writes an appropriate follow-up
3. Follows your guidelines -- every email respects your objection handling rules, never-say list, and length limits

Auto-Reply Settings

For each campaign, you can configure how replies are handled:

• Auto-send replies -- responses are sent automatically
• Auto-send delay -- add a delay (e.g., 30 minutes) before sending, giving you time to review
• Manual review -- all replies are saved as drafts for you to review before sending

Testing Your Sales Rep

Before using a sales rep in a live campaign, you can test it using the test panel on the right side of the page.

Click the chat icon in the header to open the test panel, then:

1. Write a sample email as if you were a prospect
2. See how your sales rep responds
3. Try different scenarios -- pricing questions, objections, feature requests
4. Refine settings based on the results

There are also pre-built test scenarios you can use as starting points.

Managing Multiple Sales Reps

You can create different sales reps for different purposes:

• Cold outreach rep -- optimized for first-touch emails
• Follow-up rep -- focused on nurturing warm leads
• Industry-specific reps -- trained on particular verticals or use cases

Each sales rep has its own knowledge, guidelines, and personality. Use the sidebar to switch between them, and the filter options to find specific ones quickly.

Team Collaboration

Sales reps are shared across your team. Anyone in your company can view and use any sales rep. Only the person who created one can delete it, but anyone can edit its settings or duplicate it to create their own version.

Tips for Getting the Best Results

• Start with the guided setup -- it does 80% of the work for you by reading your website, and the interview fills in the rest
• Take the interview seriously -- the more specific and honest your answers, the better your sales rep performs. The questions aren't random -- each one maps to a real piece of how your rep will communicate
• Don't rush the follow-up questions -- when the AI asks a follow-up based on something you said, that's where the deepest learning happens. Those are the moments where your rep goes from "pretty good" to "sounds like it's actually from our team"
• Watch the knowledge panel -- seeing your rep's understanding build in real time helps you spot gaps or incorrect assumptions early
• Upload real email samples -- nothing teaches tone better than examples of emails you've actually sent
• Review the knowledge summary regularly -- as your business evolves, keep your sales rep up to date using the natural language editor
• Test before going live -- use the test panel to catch issues before your sales rep starts emailing real prospects
• Keep guardrails tight -- it's better to be specific about what the sales rep shouldn't say than to leave it open-ended
• Review AI suggestions -- as campaigns run, check for field suggestions regularly. They're based on real data from your market and can meaningfully improve performance
• Run more campaigns for better suggestions -- the more reply and call data the system has, the better the intelligence and suggestions become

Next Steps

• Learn about Campaigns to start using your sales reps
• Read about Insights to understand how market intelligence improves your agents over time
• See Company for company strategy and persona intelligence