Work · 22 systems on record
Every system,by the problemit solved.
Each case opens with its headline result. Open one for the facts and the story: situation, task, action, result.
Not sure which is yours? Start with your problem
Group A · 7 cases
We need more qualified leads and meetings
A1Galax Media Flagship$150,000+ in revenue from 45 booked calls in 90 days, on about $15,000 of spend: a 10× return.
Read the storyClose
A1Galax Media Flagship
- Client
- Galax Media, a digital and performance marketing agency
- Year
- 2025
- Channels
- Email and LinkedIn
- Tools
- Smartlead, n8n, Apollo with email validation, an AI auto-reply agent, Calendly, HubSpot
- Result
- $150,000+ revenue · 45 booked calls · 90 days · about $15,000 total spend · 10× return
- Milestones on the way
- 31+ sales-qualified leads in the first 20 days; $74,000 closed by day 45
- Situation
- Outbound depended on a person researching each prospect and writing each message, so volume and quality traded off against each other.
- Task
- Build an engine that finds, researches and contacts prospects with little human effort, and keeps the personalisation real at volume.
- Action
- Automated sourcing and validation, sending, and reply handling in n8n. Every reply is sorted by intent, and an AI agent answers interested replies with two meeting slots. A four-question reasoning step keeps each message specific to its reader.
- Result
- 45 booked calls and $150,000+ in revenue within 90 days, on about $15,000 of total spend. The system kept running with minimal oversight after launch.
A2Executive coaching firm Flagship81 booked calls, 11 new clients and $55,000 in revenue on about $9,000 of spend: a 6.1× return in 120 days.
Read the storyClose
A2Executive coaching firm Flagship
- Client
- An executive and leadership coaching firm (name withheld)
- Audience
- Founders and CEOs, sales and RevOps leaders, HR and L&D, at 20–500 person companies in the US and EU
- Tools
- Smartlead, n8n, an AI auto-reply agent, Apollo, Zerobounce, Calendly, HubSpot
- Funnel
- 81 booked calls → 61 sales-qualified → 43 proposals → 11 new clients
- Result
- $55,000 revenue · about $9,000 spend · 6.1× return · $111 per qualified lead · 9.5% reply rate
- Time period
- 120 days
- Situation
- The firm needed booked calls with senior buyers at a cost it could predict.
- Task
- Produce qualified meetings across three different buyer types and show the cost of each one.
- Action
- Researched and contacted prospects with messaging written for each buyer type, automated the reply handling, and used AI replies to book meetings while interest was highest. Tracked every stage from first call to signed client.
- Result
- 81 booked calls and 11 new clients in 120 days, at $111 per qualified lead.
A3UK fintech company Flagship£400K+ closed in one quarter from 80+ qualified meetings, with zero ad spend.
Read the storyClose
A3UK fintech company Flagship
- Client
- A UK fintech company (name withheld)
- What was built
- Fully automated outbound systems and inbound AI agents
- Tools
- n8n, Apollo, Instantly, Claude API and GPT-4, REST APIs and webhooks
- Result
- £400K+ closed · 80+ qualified meetings · zero ad spend
- Time period
- One quarter
- Situation
- The company wanted a pipeline of qualified meetings and did not want to buy it through advertising.
- Task
- Build that pipeline with zero ad spend.
- Action
- Built two systems covering both directions a lead can come from. The outbound system finds and researches the right prospects, writes each first message from that research, follows up on a schedule and sorts every reply by intent. The inbound AI agents answer straight away, ask the qualifying questions and book the people who fit, so no interested lead waits for a person to be free.
- Result
- 80+ qualified meetings and £400K+ closed in one quarter, with nothing spent on ads.
A4GreatNash43 qualified meetings and 11 deals worth $50K+, at a 25% close rate against an industry average of 10–15%.
Read the storyClose
A4GreatNash
- Client
- GreatNash
- Tools
- n8n, Apollo, Instantly, Claude API and GPT-4, a CRM for booking and tracking
- Result
- 43 qualified meetings · 11 deals closed · $50K+ · 25% close rate
- Situation
- GreatNash wanted a steady flow of qualified sales meetings without cold calling.
- Task
- Build a system that generates meetings automatically and feeds them into the sales process.
- Action
- Built an outbound meeting generation system that qualified leads before they reached a calendar, so the team's time went to real opportunities.
- Result
- 11 deals from 43 meetings, a close rate about double the industry average.
Shahrukh delivered exactly what he promised — 43 qualified meetings that turned into real business. But what impressed me most was how he thinks. He didn't just set up automation, he understood our market and built a system that actually resonates with our ideal clients. He's responsive, honest about timelines, and doesn't overpromise. Working with him felt like having a strategic partner, not just a vendor.
A5Business consulting firm85 qualified leads in 45 days.
Read the storyClose
A5Business consulting firm
- Client
- An operations, finance and transformation consultancy (name withheld)
- Audience
- Owners and CEOs, COOs and CFOs at $5–150M companies
- Result
- 85 qualified leads, about 13 a week
- Time period
- 45 days
- Situation
- The consultancy needed pipeline quickly from a large, senior audience.
- Task
- Reach that audience at volume without damaging deliverability.
- Action
- Controlled the sending pace and ran three messages built on outcomes: cash and margin, throughput, and change readiness.
- Result
- 85 qualified leads in 45 days.
A6AddsSystems (Sweden) — outbound63 qualified leads in 30 days across EMEA and the Nordics.
Read the storyClose
A6AddsSystems (Sweden) — outbound
- Client
- AddsSystems, a Sweden-based premium marketing SaaS
- Audience
- Heads of Growth, Demand Gen, RevOps and product marketing at 20–500 person companies
- Deliverability
- Regional sending domains (.se and .com), SPF, DKIM and DMARC, seed tests
- Result
- 63 qualified leads, about 15 a week
- Time period
- 30 days
- Situation
- A SaaS company selling into a regional market where generic US-style outreach lands badly.
- Task
- Reach growth and RevOps leaders with mail that arrives in the inbox.
- Action
- Set up regional domains and authentication first, then ran the campaign.
- Result
- 63 qualified leads in 30 days.
A7Accounting and fractional finance firm (US)18 qualified consultations in 20 days.
Read the storyClose
A7Accounting and fractional finance firm (US)
- Client
- A US accounting and fractional finance firm (name withheld)
- Audience
- Owners and CEOs, COOs and controllers at $3–50M companies
- Result
- 18 qualified consultations, about 6 a week
- Time period
- 20 days
- Situation
- A finance firm needed consultations booked fast.
- Task
- Reach finance decision-makers with a reason to talk now.
- Action
- Three messages built on outcomes: a 13-week cash view, month-end close in under 5 days, and compliance and risk.
- Result
- 18 consultations in 20 days.
We need more qualified leads and meetings
Group B · 3 cases
We need to grow without hiring
B1Content production client Flagship$5K to $65K+ a month in about 10 months, with zero added headcount.
Read the storyClose
B1Content production client Flagship
- Client
- A content production company (name withheld)
- What was built
- A fully automated lead generation engine, GTM automations and CRM automations
- Tools
- n8n, GoHighLevel, Apollo, Instantly, Claude API and GPT-4
- Result
- $60,000 more a month · a run-rate of $720,000 a year · a 1,200% increase · no new hires
- Time period
- About 10 months
- Situation
- The company sat at about $5K a month. Finding new clients depended on someone prospecting and following up by hand, so growing meant hiring people to do more of it.
- Task
- Grow revenue without adding headcount.
- Action
- Built three systems that hand work to each other, so a prospect goes from stranger to booked call with nobody touching it.
- 01A lead generation engine. It finds companies that match the ideal client, validates the contact details, researches each one, and writes the first message from what it found.
- 02GTM automations. They send, time the follow-ups, and read every reply. Each reply is sorted by intent, and an interested reply gets an answer offering two meeting times while the interest is still fresh.
- 03CRM automations in GoHighLevel. Every lead, reply and meeting is logged and routed automatically, so the pipeline is accurate without anyone updating records by hand.
- Check
- I stayed on the system after launch, watching it and fixing it when a connected tool changed.
- Result
- $5K to $65K+ a month in about 10 months, with zero added headcount.
B2Davinchi.io Flagship30 virtual assistants down to 5, with higher output from 5 people than from 30. 83% lower labour cost, about $150,000 saved a year.
Read the storyClose
B2Davinchi.io Flagship
- Client
- Davinchi.io, social media operations
- System parts
- AI core engine, social scheduling, publishing, reporting
- Tools
- Claude API and GPT-4, n8n and Make.com, a prompt management layer
- Result
- 30 → 5 people · higher output · 83% lower labour cost · about $150,000 saved a year
- Situation
- Operations ran on 30 virtual assistants. It was expensive, and more output meant more people.
- Task
- Increase output without growing the team.
- Action
- Mapped the manual work step by step, decided which steps AI could do reliably and which needed a person, and rebuilt the operation around custom AI systems. Rule-based work runs automatically and people handle the judgement calls. A prompt management layer keeps the AI output consistent.
- Result
- Five people now outperform the original thirty, at 83% lower labour cost.
I've worked with developers and agencies before, but Shahrukh is different. He doesn't just build what you ask for — he challenges your thinking and finds better solutions. When we came to him with our VA problem, he could have just automated a few tasks. Instead, he redesigned our entire workflow and built AI systems that fundamentally changed how we operate. The guy is sharp, works fast, and genuinely cares about solving the real problem. No ego, just results.
B3AddsSystems (Sweden) — proposal systemAn automated proposal system that lets the same sales team handle more prospects.
Read the storyClose
B3AddsSystems (Sweden) — proposal system
- Client
- AddsSystems, a Sweden-based premium marketing SaaS
- What was built
- An automated proposal system
- Tools
- Make.com, Google Docs and an AI model
- Situation
- Writing proposals by hand limited how many prospects each person on the sales team could handle.
- Task
- Let the same team handle more prospects without lowering proposal quality.
- Action
- Set up a system that produces a finished, tailored proposal from a few details about the prospect.
- Result
- Proposals go out without the manual writing step.
We need to grow without hiring
Group C · 5 cases
Our operations are manual and slow
C1Mecimart (Sweden)$8K to $25K a month in 5 months (+212%), with 15+ hours of manual work removed every week.
Read the storyClose
C1Mecimart (Sweden)
- Client
- Mecimart, mecimart.se, e-commerce
- What was built
- End-to-end operations automation and a custom Shopify storefront
- Tools
- Shopify, n8n or Make.com, Klaviyo
- Result
- $8K → $25K a month · +212% · 15+ hours saved weekly
- Time period
- 5 months
- Situation
- An e-commerce business at about $8K a month with heavy manual operations.
- Task
- Grow revenue and remove the manual workload.
- Action
- Started from how the operation ran day to day, automated the repetitive work, and built a storefront that fit the business.
- Result
- $25K a month within 5 months, and 15+ hours of manual work gone every week.
C2Leenweb (Netherlands)10,000+ products updated automatically, replacing a fully manual process.
Read the storyClose
C2Leenweb (Netherlands)
- Client
- Leenweb, leenweb.nl. A Dutch product-data agency serving established brands: PIM implementation, webshops and AI product content
- Worked with
- Martijn Leenen, directly
- What I helped set up
- Two services Leenweb now sells: AI automation (images, text and translations for product catalogues) and PIM implementation (product data managed centrally and distributed to webshops, marketplaces and dealers)
- Tools
- KatanaPIM, Claude API and GPT-4, n8n or Make.com, REST APIs and webhooks
- Result
- 10,000+ products handled automatically
- Their public line
- “Enrich hundreds of products in minutes. Images, text, and translations, fully automated.”
- Situation
- Leenweb manages product data for brands with large catalogues. Product content was updated by hand across more than 10,000 products: slow, and wrong in places nobody could find.
- Task
- Replace the manual process with automation accurate enough to run on client catalogues, and turn it into something the agency could offer its own clients.
- Action
- Worked directly with the owner to set up AI enrichment of product images, text and translations, and the PIM set-up that manages product data centrally and pushes it to webshops, marketplaces and dealers. Built updates that are safe to repeat and that handle changes in the source data.
- Result
- 10,000+ products handled automatically. Leenweb now sells AI automation and PIM implementation as headline services.
Shahrukh transformed how we manage our product catalog. We were drowning in manual updates across 10,000+ products, and he built us a system that just works. What stood out was his attention to detail — he asked questions most developers wouldn't think to ask, understood our specific workflow challenges, and delivered a solution that integrated perfectly with our existing setup.
C3Mid-sized marketing firm — split-test pagesNew test pages go live in 3 hours instead of a week.
Read the storyClose
C3Mid-sized marketing firm — split-test pages
- Client
- A mid-sized marketing firm (name withheld)
- Tools
- Claude and ChatGPT models
- Result
- Split testing 10× faster · a new page built and live in 3 hours, down from about a week
- Situation
- Split testing was a bottleneck. Building and setting up every new test page took about a week, and slow tests are expensive when ad money is being spent.
- Task
- Remove page building as the limit on how fast the team could test.
- Action
- Fully automated building the pages, setting them up and putting them live.
- Check
- Every page is still reviewed by hand before it goes live, forms included.
- Result
- A week became 3 hours.
C4SverigewebbClient delivery automation for a marketing agency.
Read the storyClose
C4Sverigewebb
- Situation
- A marketing agency whose client delivery ran on manual steps.
- Task
- Automate delivery in a way that fits how the team already works.
- Action
- Learned the agency's delivery process first, then built the automation around it, and raised likely problems before they happened.
- Result
- Automation the team uses as part of its normal delivery.
Working with Shahrukh was one of the best decisions we made for our agency. He's not just technically skilled — he understands marketing strategy, which is rare. He took time to understand our client delivery process and built automation that actually fits how we work. What I appreciated most was his proactive approach — he spotted potential issues before they became problems and suggested improvements we hadn't even thought of.
C5Scoodia (Stockholm)AI workflow automation and CRM integrations for client operations, January to September 2025.
Read the storyClose
C5Scoodia (Stockholm)
- Action
- Implemented AI-driven workflow automation and CRM and data integrations, including onboarding automation and multi-tool pipelines across n8n, Make, APIs and webhooks.
Our operations are manual and slow
Group D · 7 cases
We need AI that works reliably
D1Mid-sized marketing firm — AI sales and marketing agents FlagshipOn one client account, conversion went from 8% to 34% in 3 weeks, and revenue on the account rose about 4× on the same ad spend.
Read the storyClose
D1Mid-sized marketing firm — AI sales and marketing agents Flagship
- Client
- A mid-sized marketing firm (name withheld). The work was on one of its client accounts
- Tools
- Claude and ChatGPT models
- Result
- Conversion rate 8% → 34% · about 4× revenue on the account · same ad spend
- Time period
- 3 weeks
- Status
- Still a client; the system is still being improved
- Situation
- One client account was converting only 8% of customers.
- Task
- Raise conversion without raising ad spend or adding sales staff.
- Action
- Built AI sales and marketing agents that guide each customer through every step. They learn from every conversion and are updated daily.
- Check
- People supervise the agents and each daily update. Self-improving agents need someone watching them.
- Result
- 34% conversion within 3 weeks and about 4× the revenue on the account, on the same ad spend.
Great experience working with Shah. He joined us as AI automation expert and owned our automation and AI agent systems end to end: n8n workflows, AI chatbots and CRM integrations, plus a few Python pieces where we needed them. Detail oriented, fast, and easy to work with. Already planning more work with him.
D2Mid-sized marketing firm — in-house chatbots$10,000 a month saved by replacing outside chatbot tools with an in-house team of bots.
Read the storyClose
D2Mid-sized marketing firm — in-house chatbots
- Client
- A mid-sized marketing firm (name withheld)
- Tools
- Claude Code, Claude and ChatGPT models, hosted on the company's own private servers
- Result
- $10,000 a month saved · higher conversion than the tools it replaced
- Situation
- The company was paying heavily for outside chatbot tools.
- Task
- Cut that cost without losing quality.
- Action
- Built a team of chatbots with Claude Code and hosted them on the company's own servers, so it keeps control of the bots and the conversation data.
- Check
- Quality is measured by conversion rate.
- Result
- $10,000 a month saved, with better conversion than the old tools.
D3Chat automation at scale20,000–30,000 Instagram direct messages a day handled automatically.
Read the storyClose
D3Chat automation at scale
- Agency account: 20,000–30,000 Instagram DMs a day handled through ManyChat.
- A UK-based brand: its whole Instagram outreach funnel automated, at 10,000+ DMs a day.
D4Voice agentsAI voice agents set up for Mecimart and for a restaurant in Sweden.
Read the storyClose
D4Voice agents
- Tools
- Claude API and GPT-4 for the agent, n8n for the connections
D5Great NashOne owner for the whole technical side: AI agents, internal tools and integrations.
Read the storyClose
D5Great Nash
- Situation
- The company needed AI systems, tooling and integrations owned by one person.
- Task
- Keep AI output consistent across products and remove manual work for the team.
- Action
- Built and maintain chatbots and AI agents for the product and internal operations, internal tools, a prompt management system, all major third-party integrations, and landing pages taken from design specs to production.
- Result
- One person owns the stack end to end.
D6CRM systems for 50+ B2B clientsCRM and workflow architecture for 50+ B2B clients, kept running after launch.
Read the storyClose
D6CRM systems for 50+ B2B clients
- Clients
- 50+ B2B clients across the US, EU and Middle East
- Tools
- HubSpot, Salesforce, GoHighLevel, Airtable, Zapier, Make and n8n, REST APIs and webhooks
- Situation
- CRMs and the automations around them break quietly when a connected tool changes how it sends data. Nobody notices until leads go missing.
- Task
- Give each client a CRM set-up that stays correct.
- Action
- Designed the data model, routing, scoring and sync for each client, then stayed on after launch to monitor it and fix it when something upstream changed.
- Result
- 50+ clients on CRM set-ups I designed and look after.
D7The four-question reasoning frameworkSolved batch convergence in AI-written outreach.
Read the storyClose
D7The four-question reasoning framework
- Situation
- Across many independent AI calls, outputs drift toward the same phrasing, so by email 200 the personalisation is fake.
- Task
- Keep every message distinct without a person reviewing each one.
- Action
- Designed four questions the model must reason through before it writes, which forces a prospect-specific path on every call.
- Result
- Personalisation holds across batches. It runs inside the outbound engines in Group A.
We need AI that works reliably
Send one workflow that keeps breaking, or book a 30-minute call.