Dan Rosenthal

I spent months hiring 33 workers.

But they're all AI agents.

They allow me to do the work of 10 people.

Here's how I structured my team:


↳ Executive Director: Delegates to all agent teams and reports back to me on WhatsApp or Slack.

Each agent team has a manager that picks the right worker and reports back to the director.

↳ Content Team:
Content Manager
Content Ideator
LinkedIn Post Writer
YouTube->Blog Repurposer
LinkedIn->Newsletter Repurposer
YouTube->LinkedIn Repurposer
Linkedin->X Repurposer

↳ Comms Team:
Comms Manager
Slack Assistant
LinkedIn DM Assistant
WhatsApp Assistant
Calendar Assistant
Gmail Assistant

↳ Sales Team:
Sales Manager
Pre-Call Assistant
Post-Call Assistant
Lead Researcher
CRM Assistant

↳ Outbound Team:
Outbound Manager
Intent Signal Analyst
Outbound Copywriter

↳ Research Team:
Research Manager
General Researcher
Swipe File Analyst
Clay Expert
GTM Strategist

↳ Marketing Team:
Marketing Manager
Ad Designer
Social Performance Analyst
Ad Performance Analyst

↳ Project Team:
Project Manager
Notion Assistant
GDrive Assistant

Relevance AI was the tool of choice here.

Comment “Agents“ if you want the full agent org chart.

2023: I became obsessed with Clay and knew it was going to change the game.

2026: I got that same feeling for Claude Code.

The GTM SaaS of the future will largely be operated through AI platforms instead of traditional UIs.

Many tools have already opened up that toolkit.

These are my 32 favorite MCPs to plug into Claude Code (for GTM):

- Apollo.io: 275M+ contact DB & GTM workflows.
- AI Ark: Prospecting DB with 70M+ companies.
- Findymail: Contact enrichment + validation.
- Google Workspace: Gmail + Drive + Docs.
- Notion: Company knowledge hub + PJM.
- Instantly.ai: Automated email campaigns.
- Supabase: Backend DB for internal tools.
- Ordinal: Content scheduling + analytics.
- RB2B: Website visitor deanonymization.
- HeyReach: Automated DM campaigns.
- Clay: Call Clay functions from Claude.
- Browserbase: AI browser automation.
- Customer.io: Marketing automation.
- Pinecone: Vector database for RAG.
- Firecrawl: Web scraping + crawling.
- HubSpot: Sales + marketing CRM.
- Warmly,: Website visitor waterfall.
- DiscoLike: Niche TAM mapping.
- Ocean.io: Lookalike list building.
- Apify: Store of custom scrapers.
- GitHub: Storing company OS.
- n8n: Workflow/agent builder.
- Airtable: DB for automations.
- Slack: Team + client comms.
- Salesforce: Enterprise CRM.
- TheirStack: Job openings.
- Serper: Google scraping.
- Sumble: Technographics.
- Figma: AIO design tool.
- Miro: Workflow mapping.
- Vercel: App hosting.
- Exa: AI web search.


(Bonus) 2 CLIs I’ve been loving:

- Landbase: AI list building + enrichment.
- Freckle.io: GTM orchestration.

Any tools that I missed? 👇

Workflows.io generates 1,500,000+ impressions/mo on LinkedIn.

Claude Code does 80% of the heavy-lifting.

We don’t use AI for copywriting, so these 10 Claude Code skills are focused on maximizing human creativity in the content creation process:

1️⃣ Ideation Agent

/linkedin-db → pulls top performers into our 2,000+ viral post database in Pinecone
/content-call-research → mines our call transcripts for angles based on our post frameworks
/content-ideator → curates hooks, formats, and real examples for a post based on the raw idea


2️⃣ Content Drafter

/linkedin-drafter → writes 3 drafts with bullet points + snippets from the research


3️⃣ Graphic Designer

/frontend-skill → creates UIs based on design system + MCPs
/graphic-designer → generates graphic mock-ups based on 100+ examples


4️⃣ Post QA + Scheduler

/hook-grader → references our viral hook database and framework to score the hook
/post-qa → uses the hook-grader, fact-checks the copy, and finds micro-improvements
/post-scheduler → queues the post in Ordinal with our preferred settings


5️⃣ Analytics Agent

/content-analytics → analyzes post metrics to deliver tactical reports
/content-strategist → finds content to repurpose from previous posts


I'll be sharing these skills soon.

Follow Dan Rosenthal to stay tuned.

The way humans use software has changed forever.

Claude Code is now our most used interface.

We uniquely blend software and human expertise.

We call it “AI-Native Services”.

Here’s how it works operationally:


1️⃣ Company OS in git

Our entire company dataset lives in one GitHub repo called Company OS.

What's inside:
- company/ - team, voice guide, design system, industry intel
- wiki/ - SOPs, playbooks, campaign guides
- clients/ - per-client context files
- raw/ - client calls, market research, competitor data
- plugin/ - 26 agents, 23 commands, hooks
- skills/ - 79 Claude skills

Data is constantly flowing in to keep it up-to-date.


2️⃣ Client repos

Every client gets their own private repo.

Same engineering pattern as the Company OS, just personalized to their account.

What's inside:
- Their ICP, voice guide, brand assets
- Historical campaigns and what worked
- Onboarding form data and deep research
- Slack threads, call transcripts, GDrive changes
- API/MCP connections to their revenue stack

Result: every team member has full client + company in every session.


3️⃣ Human interaction layer

We still log into some SaaS UIs, but Claude is slowly taking over.

Across 20 team members, the efficiency gain has been massive.

We try to automate as much of the admin as possible:
- Client onboarding
- Content research and ideation
- Skill tuning from team feedback
- Reply triage + sentiment routing
- Campaign launch pre-flight checks

So AI does the legwork, but humans ship.

So we can spend more time on strategy + creative GTM.


4️⃣ MCP + CLI engine

MCPs + CLIs let Claude act across our stack vs. just advise.

Some of our favorite MCPs/CLIs:

- GitHub - Company OS + client repos
- Findymail - email verification waterfall
- Google Workspace - client docs
- Airtable - automation backend
- Instantly.ai - email campaigns
- Slack - team + client comms
- Apollo.io - list + enrichment
- Notion - internal wiki + PJM
- HeyReach - DM sequences

Plus HubSpot, Browserbase, Supabase, Vercel, Figma, Stripe, Pinecone, Clay, Apify, Firecrawl, and more.

We're also migrating a ton of workflows to custom code.


5️⃣ Operationalize

As an AI-native services company, we're constantly optimizing how we work with AI and software.

Built into the system:
- Guardrails: safety hooks gate 94+ risky operations.
- PR-based governance: anyone on the team can propose a new skill, agent, or tweak as a branch.
- Workflows-engineering plugin: 26 agents, 79 skills, 23 commands auto-propagated. Agent swarms split tasks into 5-20 sub-agents.
- Self-improvement loop: n8n syncs tech stack data back into the Company OS. Pinecone stores past content + performance metrics for skills to query. Human corrections feed back in.

There isn't ever going to be a finish line, so we're building like it's a marathon.

---

The way Workflows.io operates wouldn’t be possible 1 year ago.

But we predicted this and made a big bet on AI-Native Services.

We bootstrapped from 2 to 20 people in 9 months.

Through offering 4 GTM services:

LinkedIn Content, Automated Outbound, AI RevOps, and ABM

When put together, they make up our own GTM system.

Here's the full system broken down:


1️⃣ ICP Model:

3 components: Firmographics, Technographics, Account-fit signals

Tools: Closed Won (HubSpot) → Analysis (Clay, ChatGPT, Claude)


2️⃣ TAM Map + AI Account Research:

We find relevant research points (e.g. current channels) on every account to tier them.

TAM tools: Apollo.io, Findymail, AI Ark, Clay, Ocean.io, DiscoLike

Account research tools: Claygent, ChatGPT, Apollo.io, Nooks, Claude, Exa


3️⃣ Contact Sourcing + CRM Sync:

We then find stakeholders (champions, decision makers, influencers) at every target account.

Everything is synced back to the CRM.

Tools: Apollo.io, AI Ark, Clay, BetterContact


4️⃣ Awareness Scoring:

The buying journey is tracked through 1st- and 2nd-party signals:

Identified → Aware → Interested → Considering → Selecting


5️⃣ Demand Generation:

3 main channels we think about:

Content, Outbound, Ads

Awareness score changes trigger cross-channel retargeting.


6️⃣ Sales Process:

We use AI to personalize the sales process and update the CRM automatically.

Tools: Ergo, Apollo.io, Nooks, Qwilr, HubSpot, Claude

-----

The goal is $30M with under 80 people.

This is the foundation we're building on 👇

Workflows.io is a 20-person AI-native GTM services company.

Claude Code helps us move 5x faster than some GTM teams we work with.

Here’s how we run our operations:

1️⃣ Map out high-level processes

For every process, we map out who’s involved, which tools we use, which automations/agents are relevant, and the related SOPs.

This is the blueprint for everything below.


2️⃣ Connect MCPs and CLIs to Claude Code

MCPs let Claude act inside your tools, so they power a lot of our agentic workflows.

We use 25+ incl. Slack, HubSpot, Instantly.ai, HeyReach, Apollo.io, n8n, Notion, GitHub and Supabase.


3️⃣ Build the knowledge layer (/wiki)

We built an auto-updating company brain in markdown.

Every client, SOP, transcript, campaign, etc. is documented in GitHub.


4️⃣ Build the behavior layer (/skills)

Skills turn an SOPs into an 80%+ complete task.

We have ~20 core skills, but a total of 50+ that we use.


5️⃣ Maintain an organized folder structure

CLAUDE(.) md, /wiki, /clients, /raw-context, and /.claude at the root.

Keeping this clean is a constant effort.


6️⃣ Give the team access

With strict hooks + guardrails, the whole team has access to the system.

They use it through Claude Code CLI in Cursor, or custom frontends we deploy on Vercel.


7️⃣ Deliver the outputs

Humans own the strategy, output, and every single mistake.

But Claude Code has taken 80%+ of the execution work off our plate.


---

We bet on AI-native services last year when we launched.

No one was talking about it, now it’s the hottest thing.

Luckily, we have a head start 👇

The European SaaS scene is still HIGHLY underrated.

Especially for GTM tech.

Here's a market map of Europe with the top 35 GTM companies:

My Favorites:
- HeyReach - LinkedIn Sequencer (🇲🇰)
- Findymail - Email Finder (🇫🇷)
- BetterContact - Phone Number Data (🇫🇷)
- n8n - Automation (🇩🇪)
- Fibbler - Ad Insights (🇸🇪)
- Jungler.ai - Signals (🇫🇷)
- influencers.club - Influencers Data (🇲🇰)
- Apify - Scraper (🇨🇿)

Rising:
- Twin - AI Agents (🇫🇷)
- Amplemarket - All-in-one Prospecting (🇵🇹)
- Attio - CRM (🇬🇧)
- Breakcold - CRM (🇫🇷)
- Enginy - Data Orchestration (🇪🇸)
- Searchable - AI SEO (🇬🇧)

Unicorns:
- Pipedrive - CRM (🇪🇪)
- Aircall - Dialer (🇫🇷)
- Make - Automation (🇨🇿)
- Synthesia - AI Video (🇬🇧)

Others:
- Cognism - Database (🇬🇧)
- Snitcher - Signals (🇳🇱)
- Trigify.io - Signals (🇬🇧)
- TheirStack - Buying Signals (🇬🇧)
- PhantomBuster - Data Scraping (🇫🇷)
- folk - CRM (🇫🇷)
- ZenABM - Ad Insights (🇬🇧)
- AuthoredUp - LinkedIn Analytics (🇬🇧)
- Dreamdata - Marketing Attribution (🇩🇰)
- Positive Surfer - SEO (🇵🇱)
- Dealfront - Funding Data (🇩🇪)
- Captain Data - Data Provider (🇫🇷)
- ZenRows - Data Scraping (🇪🇸)
- PredictLeads - Data Enrichment (🇸🇮)
- Teamfluence - Signals (🇪🇪)
- Crono - Sales Orchestration (🇬🇧)
- Albacross - Signals (🇸🇪)

What are the best European GTM companies I missed? 🤔

Sequoia published "Services: The New Software."

I went all-in on this idea 11 months ago.

We call it "AI-Native Services".

“The next $1T company will be a software company masquerading as a services firm.”

Sequoia’s thesis:
- For every $1 spent on software, $6 are spent on services.
e.g. Companies pay $10K/yr for Quickbooks, and $120K/yr for an accountant.
- The next legendary company will just close the books under one hood.

28% of the latest YC batch were AI-native services companies.

All types of markets e.g. insurance, supply chain, healthcare.

The service companies compounding domain data right now are building their moat.

We see a 9-figure+ opportunity in GTM.

And we're coming for it.

SF is stealing all the AI buzz.

But London is leading the charge in Europe.

These are 30 leading AI startups coming out of London:

Hackney:

PhysicsX → AI Physical Engineering
Cleo→ AI Financial Assistant
Metaview → AI Recruitment
Encord → AI Training Data
Tractable → AI Appraisals
Wordsmith AI → Legal AI
Robin AI → Legal AI

Islington:

Causaly → AI for Life Sciences
Wayve → Autonomous Driving
Signal AI → AI Risk Analysis
Faculty → AI Consulting
Oriole → AI Networking
Ori → AI Infrastructure
Attio → CRM

Camden:

Rossum → Document Processing
BenevolentAI → Drug Discovery
Fyxer AI → AI Email Assistant
Vertice → AI Procurement
Unitary → AI for Insurers
Synthesia → AI Video

Westminster:

Multiverse → AI Upskilling
ElevenLabs → Voice AI
Luminance → Legal AI
PolyAI → Voice AI
V7 → AI Agents

Tower Hamlets:

Isomorphic Labs → Drug Discovery
Granola → AI Notetaker

Lambeth:

Quantexa → Decision Intelligence

Southwark:

Monolith → Engineering Testing

Kensington & Chelsea:

Stability AI → Generative AI

—

I couldn’t capture everyone in this post.

Who did I miss?

After growing from 0→$1 mil ARR in 3 months, Workflows.io is looking to fill 4 new positions.

POSITIONS
- GTM Content Lead
- RevOps Manager
- GTM Engineer
- Content Writer

CONTEXT
After publicly launching in July, it's been the craziest months of my and Fivos' lives. We are wearing 100 different hats and had to slow down on TOFU to make sure we can continue delivering an exceptional service.

COMPANY VISION
The goal: Help the best technologies in the world find their ideal customers.
How: Become the #1 resource for GTM (services, education, products).
What: Lots upcoming, but right now we help companies through three main services: content, outbound, and RevOps

REQUIREMENTS
- You're a self-motivated builder.
- Everything is "figureoutable" mentality.
- Genuine curiosity about AI and testing new tools.

PERKS
- Learn a ton from the $1M-$10M ARR journey.
- Get paid to build your personal brand on LinkedIn.
- Time and location freedom, as long as you get stuff done.
- Future-proof your skills by embedding the newest AI tools into your work.

Apply here: https://lnkd.in/e3nbXsZt
(tag anyone who's a good fit in the comments)

Workflows.io has completed the 0→1.

We're looking for exceptional new talent to help us go from 1→10.

Highlights from the last 6 months:
- Went from $0→$1.5M ARR.
- Scaled to 13 team members.
- Booked 470+ meetings, 85% inbound.

And best of all, 100% bootstrapped.

So we've been able to set aside a significant amount of equity for anyone who comes in and makes a big impact.

4 open roles:

1️⃣ Head of Operations
We're stacking internal automations, templates, and custom-built software to the point where what we're delivering will look a lot more like "service-as-a-software" versus a pure service.

This position will become the orchestrator of this motion.

2️⃣ HubSpot & RevOps Consultant
Our revenue is evenly split between RevOps, Outbound and Content projects. The reason why we are able to cut through the noise in the RevOps world is because what we build is highly focused around GTM enablement.

This position will build the most advanced ABM systems out there.

3️⃣ GTM Engineer
The skillset required to build GTM systems and run unsiloed outbound is as high as it's ever been. We get the best results by combining volume, signals, and micro-campaigns.

This position will forge demand in markets by combining tech and sales fundamentals.

4️⃣ Content Writer
We've generated over 15,000,000+ impressions on LinkedIn for us and our clients. Not regular company page stuff. We create visually-appealing posts that educate and convert audiences 24/7.

This position will drive the most powerful B2B inbound channel out there right now.


All 4 of these roles are not for the faint-hearted.

Every single member of Workflows.io is a grinder.

And we reward that with:
- Comps that scale based on impact.
- Complete location and time-freedom (as long as work gets done).
- Enabling every single team member to build their brand on LinkedIn.

So if you're out reading this and you want to:
- Build things without cookie cutter guidance like in corporate.
- Discover the best of yourself through deep challenges.
- Stack skills for your future entrepreneurial ventures.

Then apply using with the link in the comments 👇

We're a 20-person AI-native services company and every single team member uses Claude Code + our Company OS GitHub repo.

That repo is our entire company brain: every SOP, client, campaign, and skill lives in it as markdown files.

I've continuously optimized our repo structure over the last 6 months.

Here's what I believe is the optimal setup (at least for our business):

━━━━━━━━━━━━━━━━━━━━━━━━━━

// 𝗰𝗼𝗺𝗽𝗮𝗻𝘆 𝗢𝗦/ - our repo on GitHub.

𝗖𝗟𝗔𝗨𝗗𝗘 .𝗺𝗱 - the core brain file that contextualizes the repo + high-level instructions.

𝗜𝗡𝗗𝗘𝗫 .𝗺𝗱 - content catalogue so Claude can find things efficiently.

.𝗺𝗰𝗽 .𝗷𝘀𝗼𝗻 - connects Claude to external tools. Our whole team gets 15+ MCPs with zero setup.

━━━━━━━━━━━━━━━━━━━━━━━━━━

/ .𝗰𝗹𝗮𝘂𝗱𝗲/ - this is where the core AI instructional data lives.

𝘀𝗲𝘁𝘁𝗶𝗻𝗴𝘀 .𝗷𝘀𝗼𝗻 - project config: hooks, permissions, env vars.

/𝘀𝗸𝗶𝗹𝗹𝘀/ - each skill handles one task e.g. TAM mapping, campaign building, content ideation (+ 30 more).

/𝗮𝗴𝗲𝗻𝘁𝘀/ - for cases when you need to spin up specialized sessions within a skill.

/𝗰𝗼𝗺𝗺𝗮𝗻𝗱𝘀/ - slash commands to streamline ops for your team (e.g. stringing multiple skills together).

/𝗵𝗼𝗼𝗸𝘀/ - output checks that Claude will conduct across every session.

/𝗿𝘂𝗹𝗲𝘀/ - tells Claude how to behave depending on what file it's working in.

━━━━━━━━━━━━━━━━━━━━━━━━━━

// 𝗿𝗲𝗽𝗼 𝗿𝗼𝗼𝘁/

/ 𝗰𝗼𝗺𝗽𝗮𝗻𝘆-𝘄𝗶𝗸𝗶/ - hosts all company knowledge so Claude can fully answer internal questions:

/𝗰𝗼𝗺𝗽𝗮𝗻𝘆/ - team, SOPs, ICP, offers.

/𝗺𝗮𝗿𝗸𝗲𝘁𝗶𝗻𝗴/ - voice, design, positioning, copy.

/𝗮𝘂𝘁𝗼𝗺𝗮𝘁𝗶𝗼𝗻𝘀/ - n8n/SDK docs, API docs (one .md per external API + helper scripts), infra files.

/𝘀𝗲𝗿𝘃𝗶𝗰𝗲𝘀/ - one folder with detailed info per service line.

/ 𝗰𝗹𝗶𝗲𝗻𝘁𝘀/ - one folder per client. High-level context + a pointer to their dedicated repo.

/ 𝗿𝗮𝘄-𝗰𝗼𝗻𝘁𝗲𝘅𝘁/ - the input layer:

/𝘁𝗿𝗮𝗻𝘀𝗰𝗿𝗶𝗽𝘁𝘀/ - meeting transcripts, auto-synced by category.

/𝗱𝗼𝗰𝘂𝗺𝗲𝗻𝘁𝘀/ - decks, briefs, PDFs, resources, external docs.

/ 𝗮𝗿𝗰𝗵𝗶𝘃𝗲/ - where old content goes that is too valuable to delete.

━━━━━━━━━━━━━━━━━━━━━━━━━━

// 𝗰𝗹𝗶𝗲𝗻𝘁 𝗢𝗦/ - optional: separate repos for clients in service-based or enterprise businesses

𝗜𝗡𝗦𝗧𝗥𝗨𝗖𝗧𝗜𝗢𝗡𝗦 .𝗺𝗱 - structure, SOW, preferences.

/𝗿𝗮𝘄-𝗰𝗼𝗻𝘁𝗲𝘅𝘁/ - full context from GDrive, Slack, Notion, and call transcripts.

/𝗺𝗲𝗺𝗼𝗿𝘆/ - work logs and source of truth for client delivery.

/𝗰𝗼𝗻𝗳𝗶𝗴-𝗳𝗶𝗹𝗲𝘀/ - client settings and tool setup files.

/𝗯𝘂𝗶𝗹𝗱𝘀/ - stores workflows and client-specific automations.

━━━━━━━━━━━━━━━━━━━━━━━━━━

For non-technical teams, having this level of control is the main reason why you may still want to use Claude Code as your primary AI interface.

The functionality you unlock when this is set up, is worth all the hype.

The GTM tech industry is changing quickly.

New challengers are starting to win in every sub-category.

Here are some of the trends I'm seeing:

- ABM
6sense → Seam AI

- Website Visitor Deanonymization
HubSpot → RB2B

- Data Sourcing
ZoomInfo → Clay

- Contact Enrichment
Clearbit → Findymail

- Phone Numbers
Cognism → BetterContact

- LI Outreach
Sales Nav → HeyReach.io

- Email Outreach
Intuit Mailchimp → Instantly.ai

- Workflow Builder
Zapier → n8n

- Inbound Orchestration
Chili Piper → Default

Follow Dan Rosenthal for more GTM content.

Workflows.io scaled to a 20-person team in under a year, with 0 external funding, thanks to LinkedIn content.

It's been a life-changing skill to have learned.

So I created a dense 38-minute LinkedIn content guide on YouTube.

It covers the whole system:

1. Profile Optimization
2. Content Strategy
3. Content Ideation
4. Design
5. Copywriting
6. Repurposing
7. Lead Gen & Retargeting

Link is in the pinned comment 👇