I help financial leaders generate leads with AI on LinkedIn | LinkedIn Marketing Specialist

I'm Tim Sergev, the COO at Pipeline Upgrade. We help UK IT providers fill their calendars with qualified meetings without the gamble of retainers or hiring SDRs.

We've delivered 2,500+ qualified meetings and generated £10M+ in client revenue by solving one critical problem: getting IT providers in front of buyers exactly when they're ready to switch.


HERE'S HOW WE WORK:

You tell us your ICP (company size, sector, services, buying signals). We find them, qualify them, and book the meetings. You only pay when they show up qualified.

No retainers. No ramp periods. No wasted ad spend.


WE CAN HELP YOU IF:

- You're an MSP, VAR, or IT reseller doing consistent revenue but growth has plateaued. You need predictable pipeline without hiring a sales team

- You've burned cash on retainer agencies (£5K+/month) that delivered reports instead of meetings. You want results you can actually measure

- You want 10-20 qualified meetings per month with IT buyers actively evaluating new providers


REAL RESULTS:

• One client got 80 meetings in 60 days and created a £2M pipeline. 

• Another invested £10K with us and signed a £250K contract.

• Our meeting rejection rate sits below 10% (most agencies are 30%+).


SO IF YOU:

- Are a UK IT services provider doing £10M+ annually
- Sell solutions with contract values £20K+
- Want your sales team calendar filled with qualified IT buyers

Send me a DM, or book free consultation at 👉 https://www.pipelineupgrade.com/

Tim Sergeev

Writing SOPs doesn't mean your operations are good.

It only means you wrote instructions.

And there's a big difference.

Many companies think this is operations:

→ Write a SOP
→ Share it with the team
→ Hope people follow it

That's not operations, that's documentation.

Let's take a simple example.

Imagine you run an e-commerce store.

You create a SOP for refunds.

The SOP says:

→ Customers send an email with the request
→ Someone reviews it
→ The team processes the refund

Better than nothing, but the process is still loose.

Because:

→ People can skip steps.
→ Information can be missing.
→ Nobody knows exactly where the request is.

Now compare that with a defined process.

1. Customer receives a damaged product.
2. Customer submits a claim form
3. Once a claim form is submitted
    ↳ A ticket is created automatically
    ↳ A responsible person is assigned
    ↳ The case enters a predefined workflow

Then the system guides the next steps.

→ Review
→ Approval or rejection
→ Resolution

At every stage you can see:

→ how many cases are open
→ how many are resolved
→ who is responsible
→ how long each step takes

You see the difference ;)

SOPs explain what should happen.

Whereas operations make sure it actually happens.

--- 

Tim Sergeev
COO @ Pipeline Upgrade
I make business run smoothly

COO @ Pipeline Upgrade | Helping UK MSPs Win More Deals by Giving Early Access to Verified IT Contract Renewal Opportunities

There's a difference between paying for a service and paying for a result.

It's not about pricing.

It's about risk.

Most companies sell services.

You pay the monthly fee. They deliver the activity.

→ Campaigns launched
→ Content published
→ Outreach sent

What happens after that is uncertain. Nobody guarantees results.

------

Now, let's compare that to:

“Pay X. Get Y.”

The moment the outcome is defined, the model changes.

Because someone carries the risk.

In a service model:
→ The client carries performance risk.

In a result model:
→ The provider carries performance risk.

That's why pay-by-result is harder to offer.

It requires:
1. Predictable systems.
2. Operational control.
3. Confidence in your numbers.

------

As a COO, I can tell you this:

If your internal systems are not tight, you simply cannot offer results safely.

In our case, everything is structured around delivery.

→ clear metrics
→ measurable inputs
→ controlled processes
→ transparent reporting

That's the only reason a result-based model works.

------

Now I have a question for you.

When you select a service provider

Would you rather carry the risk or work with someone who does?

Othmane Khadri

We help B2B tech SaaS startups build an acquisition funnel that generates leads and not just traffic | Generated over $800,000 in pipeline opportunities for clients | Founder Earleads

(scaling Earleads from $1M → $10M ARR in public)

If you’re reading this, you probably looked into AI SDRs or cold emailing agencies but something felt off "where's the human touch here?..." i feel you, what we do is completely different.

At Earleads we build and manage AI native go to market systems for B2B companies. AI native means we've designed them to maximize both human taste and machine's efficiency. You get best of both worlds without losing credibility in your market. 

What we do is dead simple: 

1) We get your ICP attention with social engineering content 
2) We monitor all signals shown by your ICP to engage with them when it makes sense 
3) We use targeted ADS to drive more qualified traffic toward your offers

We've worked with over 40 B2B companies helping them generate over $15 million in pipeline. We've booked over 1,500 qualified conversations and drove > 20,000 free trials accounts to our clients. 

If you want to learn more about our systems, DM me.

And if you want to follow the crazy road to $10M in ARR, i am documenting the back end of the journey on a weekly newsletter here: https://taap.it/YlAB6v

Head of Staff

I just built the playbook for GTM engineers who want to make the jump from the Clay era to the Claude Code era. It's free and you can access it right now.

Clay defined the last era of GTM engineering. The next one just started and most GTM engineers are going to miss the shift.

Clicking buttons on a UI is done.

You can now both automate execution and orchestration.

Claude Code can read your ICP, scans for buying signals, scores relevance, picks the angle, and drafts the approach. The part that used to take you 20 minutes per account now takes seconds.

GTM engineering is splitting into 2 roles:

1) Operators: They know the tools. They run playbooks someone else designed. Every junior SDR will learn this stack in 6 months.

2) Engineers: They build the systems. They encode their GTM judgment into skills that chain together and run autonomously. Signal detection feeds scoring feeds outreach feeds reporting. No manual steps.

The operators get commoditized. 

The engineers become irreplaceable.

So i packaged everything into an open-source playbook:

- 10 installable Claude Code skills: ICP building, signal detection, lead scoring, outreach sequencing, competitive battlecards, meeting prep, CRM hygiene, weekly reporting, and agent architecture planning

- 100+ curated GTM resources: every repo, every open-source release, every new MCP server dropping in the GTM space. Compiled and organized

- Ready-to-use templates: ICP frameworks, signal taxonomies, scoring rubrics, outreach sequences, enrichment waterfall patterns

- A full guide on the new GTM stack: what changed, what tools matter now, and how the pieces fit together

The skills chain together. 

Your ICP feeds your signal scanner. Your signals feed your lead scorer. Your scores feed your outreach sequences. Your outreach results feed your weekly report. 

The whole system compounds.

This is the shift. From running tools to building systems. From prompts to skills. From operator to engineer.

Free, open source, installable in 30 seconds.

Comment "GTM" below and i'll send you the link.

(we must be connected before)

7 Claude Code skills help run my entire GTM motion.

LinkedIn content. Prospect research. Meeting prep. Outreach.

All of it.

i don't re-explain context every new chat.
i don't re-paste instructions every session.
i trigger a skill and it runs.

Same methodology, same quality, every time.

Here are the 7 broken down:

(access them for free)

1- ICP Sharpener

10-question interview that pressure tests your ideal customer profile and outputs a scored doc with fit, timing, access, and intent dimensions. 

Every other skill reads this doc as context

2- Voice Builder

Analyzes your best LinkedIn posts (or those of your favorite influencers) and extracts writing patterns to produce a voice doc. 

Every post after will use these as a base.

3- Idea Sourcer

Searches what's trending in your niche and returns 5 angles matched to your voice and audience.

4- Post Repurposer

Takes any post, article, or tweet and extracts the core insight to rewrite it in your voice with your own unique pov.

5- Daily Debrief

Five minute interview about your day to extract stories, insights, hot takes as content ideas.

6- Outreach Packager

Researches each prospect to find a timely signal and writes personalized messages that reference something real. 

7- Meeting Prep Brief

One-page brief before every call about the company context, contact profiles, objection handling, smart questions

And they chain together.

ICP Sharpener → feeds Outreach Packager with scoring criteria

Voice Builder → feeds every content skill with your tone

Daily Debrief → feeds Idea Sourcer with accumulated seeds

One skill's output makes the next one smarter.

Prompts = one-time instructions. Gone next session.
Skills = repeatable methodology. Gets sharper every run.

Honestly i'm just a founder who got tired of re-explaining himself to Claude.

Skills fixed that.

i packaged all 7 into a free playbook. Open source. 

Works in Claude Code out of the box.

Comment "GTM" and i'll send it over.

(must be connected before)

I'm trying to rebuild a company (Clay) that raised >$200M+ at a $5B valuation with a $200/month Claude subscription in public.

I called the project YALC and gave myself 30 days...

6 days in here's where it stands:

1) Chat interface where you describe what you need in plain English.

2) AI that reads your query and proposes the best workflow for it.

3) Automated onboarding that scrapes your website and extracts your positioning, ICP, competitors, differentiators.

4) Editable framework so you can correct anything the AI got wrong.

5) Table engine that displays results with inline feedback (approve, reject, correct).

6) Intelligence loop that learns from every action you take.

7) Provider registry that tracks which APIs perform best per segment.

8) Human review queue for every critical decision.

9) Campaign manager that treats campaigns as hypotheses.

10) MCP server so external agents can call into your GTM data.

The whole onboarding is automated. Paste your URL, YALC reads your site, extracts everything it needs, asks you a few follow-up questions to fill the gaps, and builds your GTM framework.

Then when you ask "qualify people who liked my linkedin content past 7 days" YALC already knows your positioning.

So it'll search with best suited context.

No waterfall of tabs. No copy-pasting between tools. No starting from scratch every time.

I am building the operating system every GTM engineer would love to use.

It's gonna be open source and free (expect your API usage obviously). 

If you want to be one of the first to try it let me know in the comments.

day 6/30.

The sales industry is broken for so many reasons. Your sales team is stressed ASF but you can do something  about it: 

I didn't plan to film this. I was just listening to a depressed SDR vent. Then he said something so brutal I asked if I could hit record:

Think about 2 people for a second.

Person A (Jack) is 25. Got told this role would build character and launch a career in sales. Wakes up at 6 AM to rehearse a pitch for people who never asked to hear it. 

Makes 80 calls a day. Gets hung up on 77 times. His manager tells him those 3 conversations mean he's "getting close."

He's not getting close to anything.

Person B is a VP of Ops at a saas company. Genuinely looking for a solution to a problem she's been dealing with for months. 

She posted about it on LinkedIn this morning. 

She's in a buying cycle right now.

Jack calls Person B at 2 PM with a script that opens with "is this a good time?"

She hangs up. He logs it as "not interested." His manager tells him to call the next one.

They needed each other but the system made them enemies.

This is what kills me. 
The demand exists. 
The supply exists. 

But the delivery mechanism is so broken it turns potential buyers into people who block your number.

I built Earleads to fix the delivery mechanism: 

1- You build brand awareness so Person B already recognizes your name. 

2- You monitor signals so you know Person B is actively looking right now. 

3- You reach out with context not a script. By the time you call she picks up because she's curious not cornered.

Person A stops hating their job. 

Person B gets the solution she needed.

Sales could be healthy if done properly.

Don’t let this be your team story, book your free go to market audit: https://lnkd.in/eWwqvCdV

I got 6 hours of sleep this weekend.

Worth it.

I've been heads down in Claude Code and honestly i'm a little obsessed. I gave the thing a name so my wife would stop questioning why I'm up at 3 AM.

"Working with Henry babe. Go back to sleep."

She 100% thinks I've lost it.

But this is one of those rare windows where the people who move fast and actually ship get rewarded disproportionately. The market is paying a premium for builders who master agentic tooling and roll it into their business right now.

That's exactly why we keep giving our clients new edges on go-to-market before the rest of the market catches up.

If you're nerding out like me tell me what you've built. I'll go first:

Every morning at 8 am I have an agent that scrapes the top-performing Reddit posts from the last 24 hours in my niche, matches them against my content strategy, and drops the best ones into a Notion database with a repurposable angle already attached.

What a time to be alive…

To everyone screenshotting Anthropic hiring sales people as proof SDRs are safe:

You’re about to get hit by something called self-confirming bias.

Yes Claude Code has a steep learning curve. 

Yes enterprise sales still needs a human in the loop for now.

But look at the image I’m attaching.

It’s from Anthropic’s own 2026 report. 

Sales has one of the highest theoretical AI coverage scores and one of the lowest observed usage scores.

That gap doesn’t mean you’re safe. It means the wave hasn’t arrived yet.

Average SDRs are going to get completely crushed.

As Naval Ravikant beautifully put it: THERE’S NO DEMAND FOR AVERAGE.

AI raised the bar for everything.

How to be safe? Do more. Do better. Build complementary skills. 

I haven’t slept more than 5hours for 3 weeks now building leverage for what’s to come.

You should do the same.

Andrej Karpathy (ex-Tesla AI, OpenAI founding team) just open-sourced autoresearch. This is a game changer if you're using Claude Code:

You point an AI agent at a real LLM training setup, give it an instruction file and go to sleep.

The agent modifies the architecture autonomously by running a 5-minute training loop on a single GPU. 

It checks validation loss, decides what to keep and what to discard based on YOUR instructions and keep looping all night.

You wake up to a full experiment log and an optimized model.

How to make hyper intelligence even more...intelligent? 

What's fascinating here is running ML experiments used to require a PhD team and a compute cluster. 

Now it's a repo you can clone rn.

Let me know if you would like to receive the github repo i'll DM you the link.

This is literally the key to make your agents better, more specific, and obviously spend less tokens per task :).

Day 7/30 going after the biggest fish in the go to market industry (Clay) with a $200 Claude Caude subscription. 692 new people followed me since i started this challenge.

Thank you guys really appreciate it :).

So what was built today? 

I identified the first major blocker i can't just vibe code my way through: how to automate API selection on the fly in a safe and contextual way based on any user workflow.

Needs some great architecture to do this properly.

People were right. 

Claude Code alone is not enough...

I need to hire a CTO and 10 developers rn...

Obviously joking 10-minutes spar with AGI and Claude (It?He?She?They?) found the solution.

I can solve that through 2 API providers, will A/B test tomorrow.

What would be the most insane go to market flow you'd like to see built on the fly that would require a PDH in table management inside Clay?

I'm adding a new word to my vocabulary.

Tokensmaxing (verb): the complete obsession of any Claude Power user to squeeze every last token out of their plan.

Known side effects include:

→ Running 8 parallel chats
→ Waking up at 3AM to launch agents
→ Refreshing your usage dashboard with anxiety
→ Genuine fear of going to sleep with unused tokens

If you know, you know.

Day 8/30 building an AI native Clay alternative.

Nothing works as of today.

I worked on the qualification engine, an MCP bridge for dynamic API discovery, row piping between steps, cost tracking, and an end-to-end test route.

4 feature commits. 7 fix commits after that. 11 files changed. Zero TypeScript errors. Clean build.

And yet when I run a workflow, it still returns mock data.

The Apify actors throw 404 because the actor ID format uses a slash instead of a tilde in the API URL. Fixed that. 

Then the mock fallback was swallowing errors silently so you'd never know the real provider failed. Fixed that. 

Then the session strategy between the middleware and route handler didn't match. Fixed that. 

Then React couldn't render some of the response objects as JSX. Fixed that too.

Seven fixes. Each one revealed the next break.

The pipeline has maybe 12 points of failure between "user clicks run" and "real data appears in the table." Every single one needs to work simultaneously. 

Fixing them one at a time feels like chasing a moving target.

Which is ironic because days 1 through 7 went fast. Suspiciously fast. 

Chat interface, onboarding flow, table engine, intelligence loop, provider registry, OAuth, settings page. 

I was shipping a feature a day and feeling great about it.

But shipping features and making a system work end to end are two completely different skills (us vibecoders are just learning about that, i see you developers mocking me rn but don't worry i'll make this work).

Vibe coding is incredible at features. Describe what you want, get clean code back, wire it in. But when the entire execution chain needs to hold together across 12 files, 4 abstraction layers, and 3 external APIs, you can't just describe your way through it.

You need to understand the architecture. Not conceptually. Structurally. Where does the provider resolution actually happen. What does the execution context carry at each step. Where does the fallback trigger. What gets swallowed.

I didn't understand that well enough. And the speed of the first 7 days masked it.

Today was harder than any previous day. And it taught me more.

Next few days i'm shipping 0 new features. I'm getting what i already built to actually work as it should :)

Day 9/30 building an open source Clay alternative.

I burned the whole architecture today.

Day 8 was the wall. The pipeline had 12 points of failure. Claude was guessing API provider IDs from a hardcoded list of 8 Apify actors. 

Wrong guess = mock fallback. 

I saw fake data and thought it worked...these llms really just try to please us.

So i ripped it out.

Replaced Apify with another provider (might drop the name if it's succesfull).

It's a universal API gateway. 100+ APIs behind two endpoints. You describe what you need, it finds the right API and runs it. 

I also decided to kill the workflow builder entirely.

Users don't want configurable pipelines. They want three things:

1. Find leads
2. Enrich leads
3. Qualify leads

So that's what YALC does now. Three actions. Each one is 1-2 API calls + 1 Claude call. 

I learned today that dynamic API discovery per query doesn't scale. To actually run your entire GTM through claude code what you need is a dynamic library of tasks.

A meta-skill that sits between the user query and execution. It decomposes "find SaaS companies who raised > 10m last 6 motnhs and get the founder's emails" into two steps. Then matches each step to the best task from a curated library of pre-tested, pre-configured API calls.

Each task has a known input schema, a known output schema, and a validated API call behind it. 

Users can add their own tasks. Bring their own API keys. The library grows based on what people actually need.

YALC ships with a starter set. The community extends it.

That's tomorrow's work. The meta-skill architecture.

Happy to have found a way through yesterday's pain.

Sequoia published "Services: The New Software." 

But why the smartest minds on earth decided to stop spitting on agencies and now see it as the logical evolution of software in an ai native world? Let me unpack this for you:

Before getting into this know i wrote on the topic the first time in november 2023 and kept building in that direction, which makes my company today uniquely positioned to get the most out of this new world.

So here’s what’s actually happening in saas right now.

People are realizing they can build their own tools. Not the next Salesforce. Just the exact tool they need for their specific workflow.

Tools that requires neither onboarding nor 6-month to get a feature request built for you ;).

Tools that can finally get into hyper personalization to ensure you as a user can get the most out of them.

That's the equation saas boardrooms don't want to hear.

Because if a user gets a better result from something purpose-built for their exact needs, what are you selling them?

"But building scalable software with AI isn't possible."
Sure. Who said that was the purpose of software?

You're looking at this from the lens of you building a company. Revenue, retention, moat. But the user doesn't care about your moat. 

Users just want it to WORK.

The purpose of software is to get a user to an outcome. If a tool built for their exact needs delivers a better one, you're the middleman getting cut.

Come on.

Sequoia sees it too. Julien Bek wrote that the next trillion-dollar company will be "a software company masquerading as a services firm.".

Saas as we know it is going to die.

Not today. But i doubt we’ll get many survivors in the next 5 years.

Either people build for themselves. Or saas companies figure out how to deliver hyper-personalized output at scale (they’ll be competing with agencies like us though).

I called this in 2023. 
Now Sequoia is writing about it.

Maybe i’ll be one of those who build a trillion dollar company, who knows :) 

What’s your take on this?

Clay recently launched Clay Ads. This changes everything about how go to market works:

Until now the biggest gap in B2B revenue systems was the disconnect between your outbound data and your paid media.

Your signal engine knew exactly who was in market but your ad platform had no idea.

You'd export CSVs. Upload them manually. Watch match rates sit at 30-40%. Burn budget on customers, competitors, and people who will never buy.

Clay Ads closes that loop permanently.

Here's why we'll be rolling this out for our clients: 

→ Dynamic audience sync from Clay to LinkedIn, Meta, and Google
→ Up to 90% match rates on LinkedIn (2-4x what traditional uploads get)
→ Signal-scored accounts automatically routed to the right ad stage
→ Exclusion lists that update themselves 
→ One orchestration layer for outbound, content, and paid

This is the missing piece of the allbound system we've been building toward.

Swipe through for the full breakdown ↓

P.S: Want us to build this for your GTM? Connect with us on earleads(.)com

Clay Ads: How does it work? 

Sometimes (if sales related) please just don’t give me a mic. I tend to be slightly too polarizing…

Thank you Ridha Mami for the invite, you’re building an amzing product !

Ex-VP of Sales turned founder | Building Nudge 🫨— the AI layer that keeps deals moving | Founder @ Addvocate

EP02 drop 🎥 Deal Heat - Sales tactics from the hood
Sell to keep.

If they churn, you didn’t close a deal — you sold a paid trial.

This one’s with Feat. Othmane Khadri and it’s all about acquiring customers that stay (and why retention is a sales skill, not a CS band-aid).

https://lnkd.in/e3vCRT5z

Open video: EP02 w/ Othmane (Earleads): Revenue rules. Sell to keep—0 churn. AEs/RevOps/VP/CRO. on youtube.com

EP02 w/ Othmane (Earleads): Revenue rules. Sell to keep—0 churn. AEs/RevOps/VP/CRO.

taplio

The Battle

Tim Sergeev

Othmane Khadri

Followers

Posts

Engagements

Likes

Performance ScoreTim Sergeev's LinkedIn Performance

Performance ScoreOthmane Khadri's LinkedIn Performance

Ready to grow your LinkedIn presence?

taplio

The Battle

Tim Sergeev

Othmane Khadri

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);}Tim Sergeev's LinkedIn Performance

Performance ScoreOthmane Khadri's LinkedIn Performance

Ready to grow your LinkedIn presence?

Performance ScoreTim Sergeev's LinkedIn Performance