For the last three months, I haven’t touched my CRM.

No manual lead updates.
No “move to stage” admin.
No “create follow-up task” nonsense.

And honestly? It’s been a blast.

Because the second you stop doing CRM chores all day, you get your time (and brain) back. In my case, it was so effective that I increased my number of new conversations started by 40%… and I basically doubled my ability to run real conversations with warm leads.

That’s the entire point of an AI CRM.

Not “cool AI features.”

Time back. More convos. More sales.

But here’s the thing: even though our customers love the product, a lot of them still don’t fully understand how our AI system works (what we call Vision).

So in this AI CRM tutorial, I’ll show you step-by-step how to leverage Vision, Breakcold's AI properly to:

• automatically move contacts in your CRM

• automatically create follow-up tasks

• generate AI summaries of leads

• and ask questions to AI about your pipeline (call prep, objections, closing strategy, etc.)

This article might be a bit long and honestly a bit boring.

But I swear: if you apply this properly, it’ll become one of the best investments of your life as a sales person/founder because it gives you what everyone wants:

more time + more deals, with less chaos.

Let’s go.

What Is Vision? (The Brain of Breakcold AI CRM)

The first question I always get (on calls and in support) is:

“Ok but what exactly is Vision?”

Vision is basically the AI engine that runs inside your CRM. It does a few core things:

1) It moves contacts automatically

Vision updates automatically:

• the list where the contact should belong

• the sales pipeline stage (status)

• and the tags assigned to that contact

So you don’t have to manually drag/drop leads or keep your CRM “clean.”

2) It creates CRM tasks automatically

Vision can create follow-up tasks so hyou don’t miss deals because of lack of consistency.

This is huge because most people aren’t losing clients due to marketing.

They lose deals because of:

• forgetting to follow up

• losing track

• not knowing who to ping

• getting overwhelmed

• procrastinating

• “I’ll do it later” (you never do)

3) It generates AI summaries of leads

Every time something happens with a lead (email, LinkedIn message, meeting transcript, notes…), Vision refreshes a summary.

So you always have context.

Even if you’re late to a call (we’ve all been there).

4) You can ask AI questions about leads (“Ask Vision”)

This is like ChatGPT but for your CRM.

You can ask stuff like:

• “What are the key objections this lead raised?”

• “Summarize the last 2 meetings.”

• “What should I say to re-engage them?”

• “Analyze this deal using SPIN Selling.”

And then you can go deeper with follow-up questions.

The 3 Main Purposes of an AI CRM (The Vision Framework)

Vision exists for three reasons:

1) Auto-create follow-up tasks

So you never miss a follow-up again.

The goal is:

• you open your dashboard

• the CRM tells you what to do

• you click

• you land directly in the right thread (email / LinkedIn / WhatsApp / Telegram)

• you execute

No thinking. No admin. No “where was I again?”

Just execution.

2) Auto-move leads in CRM (Lists + Stages + Tags)

Depending on what happens with leads (replies, meetings, booking links, etc.), Vision automatically updates:

• lists

• pipeline stages

• tags

Example:

• the CRM detects a conversation started

• it moves lead to Conversation Started

• then it detects Calendly booking

• it moves lead to Booked a Call

You did nothing.

So your analytics are always accurate.

3) Provide AI summaries and deal intelligence

Vision keeps summaries always up-to-date, and lets you ask questions to the AI using:

• email threads

• LinkedIn DMs

• meeting transcripts (ex: Fireflies sync)

• notes

• custom fields

This is the “AI copilot” part.

How Vision Creates Tasks Automatically (And Why It’s Not Magic)

One of the most confusing parts for people is task generation.

So let’s make it super clear.

Vision uses a “CRM Personality” (aka AI CRM Agent) to generate tasks

A Vision personality determines:

• how often tasks are created (every 3 days, 7 days…)

• how tasks are named

• the reasoning

• the action plan

• the AI suggested messages (email / LinkedIn / WhatsApp)

You can either:

1. use a default personality template

2. create your own custom personality (recommended if you want control)

Default Personalities vs Custom Personalities

Option 1: Use a default personality/agent (zero setup)

If you want something simple:

• go to Vision settings

• choose another personality

• pick “Aggressive”, “Corporate”, etc.

• done

These default ones are designed to work without thinking.

Option 2: Create your own personality (more control)

Personally I created my own, because by default, templates create tasks for EVERYTHING.

Even if you never messaged the lead.

I didn’t want that.

So I created a custom personality like:

• follow up every 3 business days

• only create tasks if there’s already a conversation thread

• except for churn-risk customers

That way, my dashboard isn’t polluted.

AI Guidance vs AI Rules (THIS Changes Everything)

Another common confusion is:

“What’s the difference between guidance and AI rules?”

AI Guidance = soft hints

Guidance is basically advice you give to the AI.

Example:

• “This list is for inbound leads.”

• “Put people here if they replied positively.”

It’s flexible. The AI can interpret.

AI Rules = strict constraints

Rules must be followed.

Example:

• “Only put leads in this list if custom field source is email or anonymous visitor.”

• “Never create follow-up tasks if there is zero conversation thread.”

This is how you get absolute control.

How Leads Move Automatically in the CRM (The Real Setup)

This is where the AI CRM becomes insane.

The “secret” is simple: naming

Contacts move automatically based on:

1. how you name your lists

2. how you name your pipeline stages

3. how you name your tags

Example:
If your stage is called “Booked a Call”, Vision knows what to do.

If a Calendly email appears, Vision detects it, and moves the lead.

No setup required.

But… sometimes the AI doesn’t move leads exactly the way you want.

So you have 2 options:

• disable parts of Vision

• or instruct it properly (best option)

The Playground: Your AI CRM Setup Weapon

The most important tool people miss is the Playground.

Inside Vision config, you can:

• search for a lead

• run analysis

• see where Vision would move them

• see the AI’s reasoning

This is how you “train” your CRM.

And I’m not joking when I say:

in 5 to 15 minutes you can fully automate everything.

Because it’s not coding.

It’s just natural language.

You test, see reasoning, adjust prompts, test again.

Done.

My Best Tips to Make Vision Work Perfectly

Tip 1: Don’t over-prompt

Counterintuitive, but true.

Often, the best guidance is literally:

“I let you judge what’s best.”

Because the AI understands the names of your lists/stages better than you think.

Over-instructing can actually make results worse.

Tip 2: Focus on lists first

Vision first decides:

“Which list should this lead belong to?”

Once the list is correct, the sales stage is easy.

So spend your effort prompting lists, not stages.

Tip 3: Use custom fields to create powerful workflows

This is where you can do insane things like:

• “Put leads here only if custom field source = email”

• “Move customer to Retention M2 if subscribed date is older than 2 months”

This creates an AI CRM that behaves like a workflow engine… without writing automation rules everywhere.

What If You Don’t Like How Vision Moves Leads?

Then disable parts of it.

In Vision settings, you can disable:

• auto-assign tags

• auto movement

• task creation

• etc.

About tags: name them like a human would

Tags work well if they’re simple:

• US / UK / EU

• startup / agency / solopreneur

Then the AI can infer from lead data and conversation context.

But if your tags are weird or super internal, the AI won’t “guess” what you meant.

In that case: disable auto-tagging.

AI Summaries: Your Call Prep Cheat Code

Each AI lead gets a Vision summary.

It updates automatically based on:

• new conversations

• meetings

• notes

• lead field updates

My workflow is simple:

• I press Cmd+K

• type lead name or email

• open lead

• read summary

• jump on call

That’s it.

It makes you look extremely prepared, even if you’re late.

Ask Vision: ChatGPT for Your CRM (With One Big Limit)

Ask Vision lets you ask questions about a lead using everything inside the CRM.

Examples:

• “Summarize the last call.”

• “What are the lead’s pain points?”

• “What should I follow up with?”

• “Analyze the deal using SPIN Selling.”

• “Where was I weak during the call?”

The Daily AI CRM Workflow (The Whole Point)

If you implement this correctly, your daily routine becomes:

1. open dashboard

2. do tasks Vision created

3. jump into inboxes (email / LinkedIn / WhatsApp)

4. focus only on active conversations

5. sell

No CRM updates.
No pipeline admin.
No “who should I follow up?”

Your job becomes what it should have always been:

create conversations → build momentum → close deals.

Final Notes: What Vision Can’t Do Yet (But Will Soon)

Some users try to do stuff like:

• “Wait 2 days in this stage, then automatically send an email.”

That’s not available yet.

Vision V1 is about:

• automation of movement

• automation of follow-up tasks

• AI summaries + Q&A

Workflow automation (timed sequences, auto-sending, etc.) will come later.

But even right now, you can already automate 80% of the admin work that kills sales productivity.

Conclusion

If you take only one thing from this tutorial, it’s this:

An AI CRM isn’t supposed to be a fancy dashboard.

It’s supposed to remove admin work so you can do more of the only thing that matters:

conversations.

More convos = more deals.

And if Vision is set up properly, you will literally start selling more… without “working more.”

Because you finally stop wasting time doing the CRM’s job.

If you want, I can also restructure this article into:

• a full YouTube script (with hooks + storytelling)

• a shorter blog post version (1200 words)

• or a “Vision Cheat Sheet” formatted like a downloadable PDF outline