How Founder-led sales can be done right? • Founder at Sellestial • 3rd-time Founder • Don't be a stranger, follow me and reach out

Is one of your main projects for 2026 cleaning up your HubSpot data?  Want to run AI in production?

If you answered yes to anything, keep on; otherwise, stop reading this.

👉 How we are different to tackle data hygiene with your internal team only:

1. Deep Data Cleaning & Enrichment Expertise
We're data cleaning and enrichment specialists who know this problem space and its traps inside out. By cleaning and enriching 100+ HubSpot portals, we've developed a deep understanding of edge cases that first-timers don't anticipate. This is not general HubSpot admin work; it's pattern recognition from seeing hundreds of messy databases.

2. Dedicated Bandwidth + Project Discipline
Data cleaning is important but rarely urgent, which means it gets perpetually deprioritized by internal teams. We provide dedicated focus with clear project boundaries that prevent the "cleaning to infinity" trap or arbitrary abandonment that internal teams tend to fall into.

3. Production-Grade AI Expertise for Data Maturity
We've built specialized AI infrastructure designed for accuracy and reliability. We are regularly deploying 7+ different AI models for different parts of the process. We tested and evaluated over 50 LLM models over the course of the last few years, and have a comprehensive understanding of their capabilities and caveats. 

The ROI for you: Your internal teams and HubSpot admins can focus on what they do best while we handle the extremely complex, time-intensive data cleanup and/or enrichment with the AI resources your team doesn't have access to.


👉 Want to talk about Sellestial?
Shoot me a DM, and we can schedule an introductory call to discuss data.

Klemen Hrovat

HubSpot quietly dropped 21 AI agents in beta.
Not one. Not five. Twenty-one.
I went through every single one, so you don't have to.

—
👉 Social Post Agent — turns calendar events into network-optimized social content.

👉 Company Research Agent — pulls news, website, CRM data, and competitors into one pre-call brief.

👉 Deal Loss Agent — analyzes closed-lost deals to surface loss patterns by segment, rep, and product line.

👉 Call Recap Agent — turns transcripts into CRM notes, follow-up emails, and flags unanswered questions.

👉 Blog Research Agent — topic suggestions, SEO research, and draft generation inside Breeze Studio.

👉 Customer Health Agent — scores account health from calls, emails, tickets, and gives you talking points + a draft email.

👉 Cross-sell/Upsell Agent — finds expansion-ready accounts and builds tailored outreach with case studies.

👉 Sales to Marketing Feedback Agent — turns objections and deal patterns into actionable themes for Marketing.

👉 Customer Handoff Agent — assembles every conversation, deal, and ticket into one brief when accounts change owners.

👉 RFP Agent — auto-fills proposals using approved answers from past RFPs and knowledge base articles.

👉 Personalization Agent — finds highest-impact personalization opportunities and generates segment-targeted variants.

👉 Knowledge Base Agent — scans closed tickets for content gaps and writes new articles from existing rep answers.

👉 Developer Tool Testing Agent — sandbox for testing custom agent tools with step-by-step diagnostics.

👉 Video Clip Search Agent — finds trending topics, matches them to clips in your library using transcript search.

👉 Closing Agent — answers buyer questions 24/7 from your sales materials, hands off to reps when needed.

👉 Shopify Store Performance Agent — 360° store metrics with industry benchmarks and action recommendations.

👉 Brand Assistant — reviews drafts and flags tone drifts, jargon, or compliance misses against your brand guidelines.

👉 Audit Analyzer — surfaces user behavior patterns and risky activity from your HubSpot audit logs.

👉 Internal FAQ Assistant — centralizes scattered knowledge into one conversational interface with cited answers.

👉 Sales Coach Assistant — reviews call transcripts and delivers personalized coaching with actionable next steps.

👉 ICP Assistant — role-plays as your ideal customer and gives persona-aligned feedback on your copy and campaigns.
—

21 agents. All in beta. All built by HubSpot.
Some are already more polished than others. 
Some might change how you work.

Which one have you tried already? Share your experience in the comments

#HubSpot #AI #RevOps #CRM #agents

Talks about HubSpot data + AI daily @Sellestial | HubSpot Community Champion

HubSpot Solution Partner Bingo.
If you've onboarded more than 3 clients, you've hit at least 9 of these.

I won't spoil all the squares.

But if "Our data is pretty clean" doesn't trigger war flashbacks - you haven't been doing this long enough.

Free space was easy to pick.

If you got bingo on the first read - you're a real one.

P.S. Which square hit the hardest? 

#HubSpot #SolutionPartner #RevOps

You can tell how many different HubSpot admins a company has had by how many different variations of 'DO NOT USE' or 'OLD - ignore' you find in property names.

You can't create a task in HubSpot with Breeze Assistant.
But you can create it with Claude.

Did HubSpot solve one of the longest-standing campaign frustrations?

You can now manually associate contacts, deals, and tickets with campaigns.

From record pages, index pages, the campaigns tool, and even workflows.

If you've ever run a trade show, a field event, or a partner dinner and wondered "how do I tie this back to a campaign in HubSpot?" - that's how.

But here's where it gets interesting.

The attribution rules aren't one-size-fits-all.

A few things worth knowing before you start associating everything.

1) Once a contact is associated with a campaign, they stay "influenced."
Removing the association doesn't undo that. So think before you click.

2) Deals linked through a manually added contact only get attributed if they close after the contact was added. Standard attribution rules apply.

3) Deals directly associated with a campaign? Always attributed. Even if already closed.

That last one is a big deal for RevOps. It means you can retroactively tie closed revenue to campaigns that influenced it offline.

And the workflow action "Add to campaign" means you can automate all of this based on your own criteria. 

No more manual busywork at scale.

This isn't a flashy AI feature.
It's plumbing.
The kind of plumbing that makes your attribution model reflect reality.

What's the first campaign you're going to retroactively associate deals with?

p.s. Available for Marketing Hub Pro and Enterprise. Enterprise gets the extra toggle to enable or disable attribution for manually associated records.

#HubSpot #RevOps #MarketingOps

RevOps is 10% strategy, 20% systems, and 70% explaining to someone why they can't just 'add a quick property'.

HubSpot pros, how do you feel after a week-long OOO?


btw, what did I miss? Which are your favourite updates from last week?

Managing 250k+ contacts in HubSpot and you're still removing duplicates manually?

You're not behind. You're just stuck in a loop.

Here's what I see in portals at that scale:
You clean up duplicates on Monday.
By Friday, new ones are already forming.

Someone imports a list from an event.
Someone else syncs records from a second system. 
A form submission creates a contact that already exists, but with a different email.

The duplicates aren't the problem. They're a symptom.

The real problem? There's no system that understands what "clean" means to you after data gets into your HubSpot.

When you're managing two platforms, let's say HubSpot and another database, with a combined 350k+ records, the question isn't "how do I find duplicates faster?"

It's "how do I stop creating them in the first place?"

That means:
- Enriching records so you actually know what you're looking at before merging
- Building automated filters that catch bad data immediately after it enters the CRM 
- Defining match rules that account for edge cases, not just obvious duplicates
- Reviewing and fixing all the various data sources, bringing data into your HubSpot

Deduplication is the last step, not the first.

The companies that get this right don't run quarterly cleanups. They build a system that keeps the database clean between the cleanups.

Are you still running data cleanup as a project, or have you turned it into a process?

#HubSpot #DataQuality #RevOps #CRM

You can edit records in HubSpot directly from Claude.

Create contacts and other records, update properties...

I like how powerful these integrations already are.

But here's what I actually want from AI + CRM.

Earlier today, I developed a skill in Claude that helps me classify a LinkedIn contact against our ICP. 

As part of the skill, I instructed it to check HubSpot to see if the contact record already exists, so I'm not creating duplicates.

It found an existing contact record in HubSpot.
Noticed the job title was outdated.
Suggested updating it.
And we updated it.

No tab switching.
No "let me go check HubSpot real quick."
No context lost.

The job title update wasn't even the task.
It was a side effect of something bigger.

That's the difference between "AI can update your CRM" and "AI works inside your CRM while doing actual work."

I have been seriously working on and playing with the Claude <> HubSpot MCP connection over the past few weeks. 

The value isn't in the individual actions.
It's in what happens when those actions are part of a larger process.

The real unlock is when your CRM stays accurate because the system self-corrects while doing other things.

P.S. Anyone else running HubSpot through MCP? Which AI are you connecting it to - Claude or ChatGPT?

#HubSpot #MCP #RevOps #DataQuality

Your reps aren't bad at updating HubSpot.

They're just human.

Every two weeks, your sales leader needs to review the pipeline. 
Every board meeting, someone scrambles to make deals look presentable.
Every 1:1, a rep explains a deal they haven't touched in three weeks.

The problem isn't discipline. It's that manual deal hygiene costs more than anyone tracks.

One team I know lost 40+ hours in January to update deal descriptions and prepare a BoD review slide.

We built them the following process.

A deal summary agent that runs on a schedule (every two weeks, automatically), and writes structured output directly into HubSpot properties:

→ Status & Next Steps: the latest activity and blockers
→ Deal Description: a clean, readable summary for BD and leadership
→ Deal Name: standardized, consistent, no more "ACME - Follow up v3 FINAL"
→ Opportunity Potential: what the best outcome could be, based on full deal context

No manual entry.
No stale data in your pipeline.
No scrambling before the board deck.

The reps get time back.
The managers get signal, not noise.
The BoD gets clarity without a pre-call cleanup sprint.

After seeing it in production, the customer said: "It even looks better than when a human was doing it."

That's one example of how we help RevOps use AI.

The data was always there.
The system just wasn't reading it.

#AIRops #HubSpot #AI #RevOps

“AIRops”  “AI for RevOps”  “HubSpot AI”  “AI RevOps”  “AI-first”  “AI-enabled”
these and many more will be use during the Profoundly kickoff.

If any of these resonate with you, welcome.

It starts on March 12, 2026 at 11 am EST

A free, expert-led virtual event focused on how AI is changing RevOps.

Topics:
- What is AIRops
- How HubSpot AI is changing day-to-day operations
- How to use HubSpot as your context infrastructure and massively scale through purpose-built solutions and production agents
- What admins, RevOps leaders, partners, and consultants need to learn
- Real examples from practitioners

If you are:
HubSpot admin
RevOps and GTM leader
Partner or consultant
this one is for you. 

P.s. you might see me during the event, don’t miss that opportunity 🤣

"We need 100% complete data in our HubSpot"

No, you don't.

I hear this from customers all the time. Can you help us get this property to 100% completion?

Sure.
Tell me which random number I should write as "number of employees"

If that data isn't publicly available, I can give you 1,000.
I can give you 5,000.
I can give you whatever makes the chart look full.

But full isn't the same as useful and relevant.

Here's what actually happens when you force people to fill out fields where there's no data available.

We had a client with a required dropdown.
Five values.
Important property for their reporting.

They made it mandatory before a deal could move forward.

Sounds smart, right?

We ran the analysis.
80% of selections were the first value on the list.

Not because it was the right answer.
Because it was the fastest click.

Complete data.
Completely meaningless.

The real problem is timing.

You're asking a rep to choose a close-lost reason on Monday morning when they're not mentally ready to reflect on it. So they pick whatever gets them past the gate.

Maybe they would've given you a thoughtful answer Friday afternoon when they'd set aside 30 minutes to clean up their pipeline.
But you didn't let them choose when.

Here's how I think about all this.

Not every property deserves the same completeness standard.

No contact should be assigned to a deal without a job title.
That's reasonable. 

But requiring job titles for every newsletter subscriber?
There's no point.

A contact entering your CRM doesn't need the same data depth as one that's a week from closing.

The goal isn't 100% complete. 

The goal is
the right data
at the right stage
for the right reason.

Sparse but relevant beats complete but useless. Every time.

P.S. Do you have a property in your HubSpot that you suspect is generating useless data right now?

#HubSpot #RevOps #DataQuality #CRM

HubSpot's Data Quality duplicate tool finds the obvious ones.

Exact same email, domain, company name.

But these aren't the duplicates that typically bother you.

I had a call today with a RevOps lead managing 14,000 companies. 
She tried HubSpot's native tool to find duplicates.
It caught some - the easy ones.

The hard ones?

A company listed as .com and .de.
Same name, but one has "GmbH" appended.
Two records that look different but are the same business.

It doesn't know.
And honestly - neither can you, without opening both websites.

That's the gap.

To resolve real duplicates, you need data outside your HubSpot.

Is .de redirecting to .com?
Is the "GmbH" version a subsidiary or the same entity?
Is the LinkedIn company page the same for both?

This is where AI agents change the game.

Not the "click a button and magic happens" kind.
The kind that opens both websites, checks redirects, compares LinkedIn pages, and gives you a confidence score before merging anything.

The data inside HubSpot indicates that there might be a duplicate.
The data outside HubSpot helps you decide if it actually is a duplicate.

P.S. What's your experience with the HubSpot Data Quality tool to help you with deduplication? 
#HubSpot #DataQuality #RevOps #deduplication

Stop forcing your sales team to enter clean data.

They won't. And even if they try, it won't be clean.

We work inside HubSpot portals every day.
We see what happens when admins add validation rules, required fields, and dropdowns to "ensure data completeness".

Here's what actually happens:
Sales reps are mid-conversation with a customer.
They need to log a job title in HubSpot.
The process demands a perfectly formatted entry.
They're not thinking about your reporting needs.
They're thinking about closing the deal.

So they type whatever gets them through the fastest.

Or worse, you make it a dropdown.
And 80% of your team picks the first value on the list.

You get complete data.
Completely useless data.

Here's what we suggest doing instead.

Let users enter messy data. Typos and all.
"Chief Tech Officer"
"CTO."
"C.T.O."
"Chief Technology Offcer"
all fine

Let them type what feels the easiest in the moment.

Then let the system standardize it behind the scenes.

A month ago, it clicked for me why HubSpot's default country/state property is a free text field.

Not a dropdown.
Free text.

Because data comes in from integrations, imports, forms, and manual edits.
Every source in a different format.

US
U.S.
USA
United States
United States of America

All valid. All messy. All cleanable.

The job of RevOps isn't to be the data police.

It's to build infrastructure that turns messy input into trusted output.

Your sales team shouldn't be responsible for typo correctness.

They should be responsible for having conversations that close deals.

You take the burden of standardization.
That's your job.

P.S. What's one property in your HubSpot that is causing you a "standardization headache"?

#HubSpot #RevOps #DataQuality #DataIntegrity

🎙️ New to streaming or looking to level up? Check out StreamYard and get $10 discount! 😍 https://lnkd.in/dAaAJzce

Email throttling has been standard practice in cold outreach for years.
Now it's coming to HubSpot marketing emails.

I've seen what happens when companies migrate to HubSpot and start sending at full speed on day one. 
Bounce rates spike. 
Spam filters trigger. 
Deliverability tanks before the first campaign even finishes.

HubSpot just released a beta called Throttle Marketing Email
and it's a meaningful addition for anyone sending marketing emails via HubSpot.

Here's what it does
When scheduling a marketing email, you can now set a custom send rate under Advanced Settings. 

Instead of blasting your entire list as fast as possible, the system spreads delivery over time at the speed you choose.

Why this matters more than most people think:
- Sending 50,000 emails in 10 minutes looks nothing like normal human behavior to inbox providers. They notice.
- Large volumes hitting a single company domain at once can quietly destroy your reputation with that domain's ESP.
- High bounce rates hurt way more when they arrive in a concentrated burst. Spread them out, and the damage is manageable.

This beta adds another layer of control to that process. 
It's not just about how many emails you send. 
It's about how fast they arrive.
And if they arrive at all.

Deliverability is somewhat a velocity problem.

Available for Marketing Hub Pro and Enterprise.

P.S. Have you tried the beta yet? Did you wish you could set this in the past?

#HubSpot #EmailDeliverability #MarketingOps #RevOps #Marketing

Unified Revenue View part 2

🎙️ New to streaming or looking to level up? Check out StreamYard and get $10 discount! 😍 https://lnkd.in/dAaAJzce

Today marks a day. For the first time, when I opened my laptop, I went to Claude Code and not to Slack, LinkedIn, or Email 🤯

Dharmesh Shah published his poll results asking his newsletter subscribers: Have you #UninstalledChatGPT and moved to Claude? 

~70% of respondents have switched or are considering switching to Claude.

The results were aligned with what we saw in the chat during the Profoundly kickoff last week. 

I have a follow-up question to this. Do you have Claude Max plan?

I help B2B Startups take their products from idea to revenue in weeks not months | Fractional Product Growth & B2B GTM Expert 🛠️ | Get a free 1:1 audit ☎️ | Founder of Buska.io 🔍

AI will not replace your team. Companies that use AI well will outpace you.

In the last two years advising B2B companies, I saw the same pattern :
- Feature-led messaging and generic AI marketing content.
- Theoretical trainings no one uses.
- AI pilots launched for FOMO. no impact on revenue.
- Random tests with AI tools. no GTM strategy behind them.

I chose a different path. I build AI-powered GTM strategies that turn clarity into pipeline.

What this means
• Strategy first. ICP. value narrative. segmentation. positioning. messaging.
• 90 day Go To Market plans your team can run. clear owners and cadence.
• AI is used as a force multiplier. faster research. sharper targeting. smarter testing. cleaner synthesis.

Who we work with
Founders and leaders in startups/scaleups who want a plan they can ship in weeks, using AI.
Mid-market and enterprise teams that need AI enablement adoption. not theory or tool demos.

We have been trusted by companies worldwide:
L’Oréal. Essilor. IGN. Mission French Tech. La Growth Machine. Taiyō.AI. Zeliq. Numixs. Ascend. Giro Music.

How I help
AI-Powered GTM strategy
Hands-on AI Enablement for C-Level.
AI trainings for teams to build adoption.

Teams always leave with concrete, actionable deliverables. 

Why this works
I sit across Marketing. Product. Marketing. and Sales.
I mix inbound content and outbound expertise to build growth flywheels.
I simplify decisions. Then I use AI to speed up heavy tasks and scale what works.

If you are done experimenting and want a GTM that gets customers fast. let’s talk.
Book a 20-minute diagnosis. https://cal.com/tonidossantos/audit

What drives me?
I want to see companies grow their business with AI, not talk about it.
I get a kick out of seeing people go from “not sure where to start” to “here’s how we’ll win the market.”

Toni Dos Santos

Sonnet 4.6 is out. Grok 4.20 is out.
Get ready to drown in "Game Changer", "AI runs my business" and "ChatGPT Killer" posts.

Meanwhile, here is what 90% of people are actually doing with the most powerful technology in human history:

- "How many eggs go in a sponge cake?"
- "Generate a realistic image of a cat playing jazz piano."
- "My boyfriend left me on read. What does it mean?"
- "Write me a LinkedIn post. Make it viral."

We are handing out Ferraris to people who can barely take a car out of the garage.

The bottleneck isn't the model context window. It's not the reasoning capability.

The technology is ready. People are not.
----
PS: If you’re asking AI for relationship advice, the hallucinations are the least of your problems.

Make AI Work for Your Growth | AI Trainer for Teams | Hands-on AI Enablement & GTM Strategy

Monday is here!  Top 6 AI news you need to read. 
I track all the AI releases so you don't have to.

1- Claude Sonnet 4.6 hits 1M context. 
Anthropic just released a cheaper brain for long, messy codebases. It actually resists prompt injection better. 
Good news for developers tired of context limits.

2- Grok 4.2 is out with four agents debating. 
Specialized bots now argue over your prompt in parallel. T
hey hand you a single answer after fighting it out.

3- NotebookLM fixes your slide decks. 
Stop rebuilding presentations from scratch. 
You can prompt your edits and export straight to PowerPoint. Save hours of busywork.

4- Google Lyria 3 makes full tracks. 
The new model generates high-fidelity music, vocals, and lyrics. 
You build them directly from text or uploaded images. 

5- Gemini 3.1 Pro levels up. Google brings a massive reasoning jump for harder tasks. It scores high in benchmarks. A step up for complex workflows.

6- Claude hunts vulnerabilities for you. 
Claude Code scans your entire codebase and suggests patches. 
You just review and approve. 

Another week, more models. 
But the real shift is clearly "AI now does the work." 

So, which update caught your attention?
----
PS: Share this with your colleague who still uses AI to "summarize this PDF."

taplio

The Battle

Klemen Hrovat

Toni Dos Santos

Followers

Posts

Engagements

Likes

Performance ScoreKlemen Hrovat's LinkedIn Performance

Performance ScoreToni Dos Santos's LinkedIn Performance

Ready to grow your LinkedIn presence?

taplio

The Battle

Klemen Hrovat

Toni Dos Santos

Followers

Posts

Engagements

Likes

Performance Score.css-goub6s{font-size:var(--chakra-fontSizes-md);font-weight:500;color:var(--chakra-colors-gray-600);display:block;margin-top:var(--chakra-space-1);}Klemen Hrovat's LinkedIn Performance

Performance ScoreToni Dos Santos's LinkedIn Performance

Ready to grow your LinkedIn presence?

Performance ScoreKlemen Hrovat's LinkedIn Performance