What does Jachin Saikia Sonowal do?

Jachin is a Growth Systems Architect who builds automated revenue engines using AI agents and marketing operations. He is the founder of PromptMatrix and Matrix Labs.

How to contact Jachin Saikia Sonowal?

You can reach Jachin via LinkedIn or email at jachinchsaikiasonowal@gmail.com. He is open to remote full-time roles in Marketing Ops and AI Automation.

Jachin Saikia Sonowal — Growth Systems Architect · Marketing RevOps · AI Marketing Automation

Timeline

THE
JOURNEY

Eight years of building growth systems under real pressure, with real revenue on the line. Every chapter is measurable. Click any to expand.

THE FOUNDATIONS ERA

Freelance & Self-Built Skills · 2016 – 2019

Pre-Job · Learning by Doing

Before any formal role, I started upskilling through digital media. During and after my MCA, I took on whatever client I could find — small local businesses, personal brand projects, digital presence builds. I taught myself SEO by ranking real websites, email marketing by running real lists, and content by producing it daily.

What I Was Building

Local business websites on WordPress — learned HTML/CSS, UI, and web performance hands-on
First Google Ads campaigns for a local education institute — ₹1.5K budget, tracked every click to enrollment
Social content for a handful of Assam-based SMEs — began understanding audience behaviour and organic reach
Built my first email list from scratch — 200 subscribers, zero budget, manually written sequences
Completed foundational certifications in Google Ads, Analytics, and digital marketing fundamentals

What This Era Taught Me

Clients don't care about traffic — they care about revenue. Learned this the hard way with early vanity reporting
Technical skills compound. Every hour learning HTML, GA, or Python paid dividends in every role that followed
Scrappy resourcefulness — doing everything yourself before you can delegate anything is irreplaceable training

MCA 2019

Gauhati University

2yrs

Self-taught hustle

1st

Paying clients

DIGITAL MARKETING EXECUTIVE

SUN Enterprise · Feb 2018 – Mar 2020

First Corporate Role · Foundation

My first professional role in digital marketing. Hired at SUN Enterprise (Guwahati-based), I quickly expanded responsibilities across the full marketing stack. Covered social media management, content creation, on-page SEO, and campaign performance reporting. This role built the practical foundation—client communication, tool proficiency, campaign discipline—that led directly to launching independent consulting in 2021.

What I Did

Daily social media management: content calendars, community engagement, audience growth across platforms
Content creation: blog articles, email copy, marketing materials, social media content
On-page SEO: keyword research, content optimization, technical improvements across 20+ pages; 40% organic traffic lift
Campaign execution: multi-channel campaigns (email, social, paid, organic) from strategy to launch
Performance reporting: weekly/monthly dashboards, client presentations, metric tracking and analysis
Team leadership: trained 2 junior team members on SEO and content best practices

What I Learned

Professional execution: delivering campaigns at enterprise scale in a team environment
Client communication: managing expectations, difficult conversations, staying calm under pressure
Tool mastery: Google Ads, Google Analytics, WordPress, email platforms, social management tools
Campaign discipline: documentation, process, consistency, and measurement—foundations for everything after
Cross-functional collaboration: building credibility while working with design, sales, and customer service teams

Why I Left (Intentional Move)

After 2 years, I had the professional foundation and confidence to work independently
The COVID-19 lockdown halted agency work, creating the pivotal window to transition fully to independent consulting.
Launched independent consulting August 2021 — a deliberate, planned pivot after building a solid foundation at SUN Enterprise.

260%

Social media growth

Performance marketing

40%

Organic traffic lift

Team members trained

VISUAL DESIGNER

The New Studio · Feb 2021 – Present (Part-time / Project Basis)

Job Role · Design & Digital

Designing digital products and marketing materials across Figma, Canva, and Creative Suite. Working on UI mockups, website carousels, landing pages, and full product visuals for mobile and web applications — giving me a visual design foundation that sharpened every marketing asset I produce.

Scope of Work

Website carousels and social templates for service businesses — professional services, retail, manufacturers
Mockups — wireframes to high-fidelity Figma mocks for local projects
Landing page designs — structure, hierarchy, CTA placement, conversion-first layout thinking
Site builds — custom themes, page builders, and performance optimisation
Marketing one-pagers, pitch decks, and proposal decks for SME clients

Design Principles I Carry Forward

Every pixel has a reason — good design is invisible, bad design kills conversion
Visual hierarchy is also information architecture — structure leads the eye toward action
Speed matters: deliver fast, iterate faster. Clients trust speed-plus-quality over perfection-plus-delay

GROWTH CONSULTANT

Independent Practice · Feb 2021 – Present · India / Remote

Independent Consulting · 175+ Engagements

175+ engagements with 120+ unique clients across B2B manufacturing, D2C, retail, professionals, political campaigns, and government organisations. Working independently across the full growth stack — strategy, execution, analytics, and team leadership.

D2C Brand Launch — ₹1.2 Crore in 18 Months

Full GTM from zero — market research, brand positioning, Shopify store build, payment gateway, logistics setup
Technical SEO + content strategy: ranked for 80+ keywords, #6 from #0 for 12 high-intent search terms
Email funnel: 0 → 2,000+ subscribers, automated welcome, nurture, re-engagement, and post-purchase sequences
Result: ₹1.2 Cr ($145K+) annual revenue in 18 months from a standing start.

Multi-Location Retail SEO — 8 Locations

Google Business Profile audit and complete rebuild across 8 physical locations
Citation cleanup: removed 40+ duplicate and inconsistent NAP listings across directories
Review generation system and staff training for 40+ frontline employees on reputation management
Result: 3× visitor growth, 280+ new reviews, star rating 3.8 → 4.6, Top-3 local pack rankings held 12+ months post-engagement

Political Campaign — 45-Day Election Cycle

Led and managed 8-person content team through an intense 45-day state election campaign
300+ content pieces produced — social posts, videos, graphics, manifestos, press releases. Zero compliance violations
Real-time content calendar adjusted daily based on news cycle and opponent activity
Result: 5× reach (50K → 2.5M+ impressions), 35,000+ new followers in 45 days

B2B Manufacturing & Healthcare

Content strategy and lead generation for a manufacturing client — 40+ qualified inbound leads in 6 months
Healthcare clinic: Google Ads + SEO combination — 60% increase in appointment bookings, maintained 4.8-star GBP rating
Government project compliance: marketing materials and reporting aligned with regulatory requirements across 3 engagements

$145K

Single client revenue

3×

Retail visitor growth

2.5M+

Campaign impressions

LEAD OPERATOR — MATRIX LABS

Newsletter & Media · Jan 2024 – Present

Solo Builder · 300+ Subscribers · 0 Ad Spend

Built and operate an AI-focused newsletter from scratch with zero paid acquisition. Matrix Labs Newsletter covers the AI and tech landscape for an Indian and South Asian audience — curated weekly, written to be actionable, not just informative. Full solo ownership: editorial, growth, monetisation, and automation.

Growth Engine — Zero to 300+

Content-led SEO and organic social distribution — no paid acquisition used at any point
Referral program design with Beehiiv built-in mechanics — subscriber-to-subscriber virality loop
Cross-promotion with complementary newsletters and LinkedIn thought leadership
Launched with a landing page converting at 8.5% — validated before building content backlog

Retention & Engagement — 38% Open Rate

50+ A/B tests over 18 months: subject lines, send times, content format, CTA placement
Open rate grew from 22% → 38% — nearly 2× the industry average of ~21%
Segmented the list by interest tag (AI tools / AI business / AI technical) for personalised content
Re-engagement sequence recovered 18% of subscribers who hadn't opened in 30+ days

Monetisation — Organic $3.8K Revenue

Validated product-market fit via 120+ subscriber surveys before writing a single word of sales copy
Built the Mega Prompt Bundle (3,500+ prompts) — validated demand before building
$3,800 in 8 months, 100+ customers across 7 countries — 100% organic, newsletter-driven
Full funnel: newsletter → landing page (8.5% CVR) → email nurture (60% open) → purchase (1.7% conversion)

$3.8K

Revenue · 0 ad spend

300+

Subscribers · 7 countries

38%

Open rate vs 21% avg

DIGITAL PRODUCT — MEGA PROMPT BUNDLE

Solo Product Creator · Mid 2024 – Present

Validated · Launched · $3.8K Revenue

Before writing a single word of the product, I ran a 120-subscriber survey campaign to validate demand. The result: a 3,500-prompt bundle covering marketing, sales, operations, and content — priced at $29, sold to 100+ customers across 7 countries in 8 months. Built entirely through newsletter-driven organic marketing.

Validation-First Product Development

120+ survey responses collected through the Matrix Labs newsletter — specific pain points mapped before any building
Three pricing anchors tested: $19, $29, $49 — $29 hit the sweet spot at highest volume × value
Beta list built before launch: 80+ pre-sale expressions of interest

Product Build

3,500 prompts across 40+ marketing and business use cases — each tested and refined, not bulk-generated
Organised by role (marketer, founder, operator, creator) and workflow (ideation → execution → reporting)
Supplementary: prompt writing framework, variable injection guide, and model selection guide included

Distribution & Sales

Digital product storefront (matrixlabs.systeme.io) — SEO-optimised listing, A/B tested cover design and copy
5-email launch sequence to newsletter list — 60% open rate on launch email, 1.7% overall conversion
Affiliate program launched post-90 days — 12 active affiliates driving secondary sales
Customer support: AI agent via Gumroad API handling 90%+ of queries without human input

$3.8K

8-month revenue

100+

Customers · 7 countries

8.5%

Landing page CVR

AI AGENT ENGINEER

Personal AI Lab · Oct 2024 – Present

25+ Agents in Production · 25hrs → 8hrs Weekly Ops

Progressed from no-code workflow automations to full AI agent development using OpenClaw (proprietary agent orchestration framework) and LLM APIs. 10+ custom agents running in live production, replacing manual marketing workflows with intelligent, event-driven systems — cutting weekly ops time from 25 hours to under 8.

Phase 1 — Workflow Automation (Oct 2024 – Jan 2025)

Newsletter pipeline: RSS ingestion → relevance scoring → editorial digest (2hrs → 15min daily)
Subscriber lifecycle: welcome sequences, engagement tags, win-back flows via Beehiiv + ActiveCampaign
Social distribution: newsletter → LinkedIn, X, Medium via API — one input, four platform-native outputs
GA4 → Looker Studio → Notion auto-report delivered every Monday with zero manual work

Phase 2 — OpenClaw Agent Suite (Jan 2025 – Present)

Content Research Agent — monitors 30+ sources, scores relevance, delivers pre-written summaries ready to edit
Newsletter Drafting Agent — research → outline → draft pipeline, 60% writing time reduction
SEO Content Agent — generates posts, auto-publishes to Ghost, submits URLs to Search Console
Subscriber Onboarding Agent — personalises welcome, infers segment, enrolls in correct nurture track
Product Promotion Agent — lifecycle-triggered, context-aware promotion (+40% conversion vs. broadcast)
Social Distribution Agent — 1 input → 7 platform-native outputs, auto-scheduled at optimal times
Customer Support Agent — Gumroad API integration, 90%+ queries resolved without human escalation
A/B Test Agent — generates subject line variants, monitors statistical significance, promotes winner automatically

~25→8hrs

Weekly ops time

22→38%

Email open rate

+40%

Product conversion

BUILDER — PROMPTMATRIX (SIDE PROJECT)

SaaS Developer Tool · Jan 2026 – Present · Beta

Building in Public · Open Source · Local-First · MIT Licensed

I was running 25+ AI agents and every prompt was hardcoded somewhere — config files, Python strings, agent briefings scattered across markdown files. Every time a briefing needed tuning, it meant finding the string, editing it, restarting the process. No version history. No rollback. No visibility. I built PROMPTMATRIX to fix that — solo, bootstrapped, starting from browser localStorage and building toward a full cloud backend.

What PROMPTMATRIX Does

Prompt registry — every agent prompt in one governed dashboard, versioned, with full history and one-click rollback
Approval workflow — propose a change, review the diff, approve → live in seconds. Nothing reaches production without a human decision
Serve endpoint — agents call /pm/serve/:key at runtime; always returns the current approved version. No redeploy ever again for that prompt
Storage ladder — browser localStorage (zero setup), SQLite (persistent, offline), self-hosted database, or managed cloud. No forced vendor, no lock-in
OpenClaw integration — built to work with my own agent orchestration framework; compatible with LangChain, AutoGen, or any raw LLM API

The Build Strategy

Open source core (MIT) ships first — build trust and GitHub presence before charging anyone
Cloud backend triggers when 5 people register — no infra cost before there's real demand signal
Building the backend live on Instagram — every architecture decision, every commit, every problem is content
Matrix Labs newsletter (300+ subscribers, 38% open rate) is the primary distribution channel — zero paid acquisition
Founding access: $149 one-time, rate-locked forever. Cloud ships at 5 registrations.

MIT

Open source license

25+

Agents using it daily

$149

Founding access · rate-locked

Services

HOW
I WORK

DIAGNOSE

Map the growth system. Identify the one constraint that explains every other symptom. Decide what to fix, what to ignore, and whether a build or a retainer makes sense. You leave with clarity — not a deck.

BUILD

Design and implement revenue infrastructure: funnel logic, CRM automation, AI agents, attribution, and an experimentation system. Fixed scope, 4–6 weeks. Ships and runs without me when it's done.

COMPOUND

Stay on as fractional owner. Refine signals, evolve agents, harden automation, run experiments. Growth stops resetting every quarter because the system — and the learning — persists.

● Open

Full-Time

Growth Marketing Lead · Marketing Ops · AI Automation & Systems

Actively applying for full-time remote roles. You get a senior operator who ships — not a strategist who decks. I own the full growth stack: acquisition, automation, AI systems, and team build-out. 8+ years, ₹1.2 Crore D2C, 175+ clients. Available immediately.

Timezone — IST (UTC+5:30) · 4hr overlap with EU · async-friendly with US

Notice — Available immediately · remote-first

Best fit — Series A–C SaaS · D2C · AI-native startups

Let's talk

◑ 1 Slot Left

Fractional

Embedded Growth Lead · 2–4 Days/Week

You get the operator, not the deck. I embed inside your team, run the growth function, and hand off a working system — not a slide with recommendations.

Commitment — 2–4 days/week · min 3-month engagement

Timezone — IST · overlap hours agreed upfront

Best fit — Scaling startups · pre-hire growth lead gap

Check availability

◈ Project-Based

Consulting

Fixed-Scope · Audit · Strategy · Build

Defined deliverable, defined end date, defined price. Audits, growth strategy docs, stack setup, AI workflow builds. No retainer that never ends.

Duration — 1–8 weeks · scope agreed before start

Remote — 100% async-compatible · all timezones

Deliverables — SEO audit · funnel strategy · AI stack build

Start a project

● Taking Builds

AI Build

Custom Agent Systems · Spec to Production

I spec, architect, and ship custom AI agent systems. No no-code wrappers. Real pipelines: n8n, Python, LLM APIs, feedback loops. 25+ agents built and open sourced.

Timeline — MVP in 2–4 weeks · full system 4–10 weeks

Stack — n8n · Make · Python · GPT-4o · Claude API

Use cases — Lead scoring · content pipeline · newsletter AI

Spec your build

Engagement Tiers

Growth Systems Diagnostic

5–7 day deep dive. I audit your current tech stack, CRM, workflows, and funnel before we speak. I map where the revenue is leaking, where agents can replace manual effort, and what to fix first. Delivered as a written system map — one primary bottleneck, a prioritised action list, and an hours-reclaimed vs. revenue-gained projection. You leave with a blueprint, not a deck.

Fixed · ₹45,000 – ₹1,50,000 · (~$530 – $1,800)
Credited toward any build within 14 days
Book the Audit →

AI Revenue System Build

Fixed-scope, 4–6 weeks. I design and implement the revenue infrastructure: funnel and lifecycle logic, CRM automation, AI agents, attribution, and an experimentation framework. You walk away with a system that runs without constant human effort — replacing the coordination cost of 1–2 hires.

Typical investment starts from ₹3,50,000 (~$4,150)
Scope and final pricing confirmed on call
Discuss Scope →

Ongoing Growth Systems Retainer

Fractional ownership, weekly cadence. I own the growth infrastructure over time — refining signals, evolving agents, running experiments, removing manual dependencies. Growth stops resetting every quarter because the system and the learning both persist.

Typical investment starts from ₹1,25,000/month (~$1,500/month)
Scope and final pricing confirmed on call
Explore Retainer →

Final pricing depends on scope, timeline, and client location. All engagements are scoped on a call — no surprise invoices. INR rates apply for India-based clients. USD for international.

The Work

CASE
STUDIES

Seven engagements. Real clients, real constraints, real revenue on the line. Each one is the full story — the problem, the decision, the execution, and the number at the end. No hypotheticals. Click to open.

D2C · Growth Marketing · Tea Industry

Zero Online Presence to ₹1.2 Crore in 18 Months — No Agency, No Team

₹1.2 Crore in 18 months. No agency. No team. One operator, one system, built from nothing.

Shopify SEO Email Marketing 18 Months Assam, India

$145K+

Revenue Generated

5K+

Email Subscribers

The Problem

A premium Assam tea manufacturer. Established offline, invisible online. Real differentiation — single-estate, artisanal — but zero digital presence, no direct sales channel, and total dependence on B2B wholesale at margins that couldn't sustain the business long-term. The brief: build a D2C channel that generates meaningful revenue within 18 months.

The Strategy

Three phases: foundation (brand, store, SEO baseline), acquisition (organic search, content, email list), monetisation (funnel optimisation, retention). I bet entirely on owned channels — search and email. No paid ads. Channels that compound over time and don't die when the budget does.

The Outcome

₹1.2 Crore ($145K+) annual online revenue by month 18. #1 for 12 high-intent keywords. Top-5 for 80+ terms. A 5,000-subscriber email list that became a predictable monthly revenue channel — still growing after the engagement ended. Not a rented audience. A compounding asset.

Execution Roadmap

Foundation

Brand positioning, Shopify build, product photography brief, payment + logistics setup. Weeks 1–4.

SEO Architecture

Keyword research (800+ terms mapped), technical SEO audit, content cluster strategy, schema markup.

Content Engine

Monthly blog cadence, 40+ product and category pages written, 12 pillar pieces targeting main buyer intents.

Email Funnel

Lead magnet, pop-up testing (A/B'd to 4.2% opt-in), 7-email welcome sequence, post-purchase, win-back.

Revenue Ops

Klaviyo flows, monthly broadcast cadence, subscription box offer, repeat purchase rate tracking + optimisation.

$145K+

Annual Revenue · 18 Months

5K+

Email Subscribers Built

#1 Rank

12 Target Keywords

80+

Keywords in Top 10

"Jachin transformed our business model. We went from zero online presence to generating over ₹1 crore in online revenue in 18 months. His systematic approach created a sustainable channel that keeps growing."

Managing Director

Tea Manufacturing Company, Assam

Political Campaign · Content Ops · Team Leadership

50,000 to 2.5 Million Impressions in 45 Days — Eight People, Zero Violations

8-person team. 45 days. 300+ pieces. Zero violations. 50K to 2.5M+ impressions.

Team Lead Social Strategy Content Ops 45 Days Zero Violations

2.5M+

Impressions

5×

Reach Growth

The Problem

A political campaign. 45-day window, strict compliance, 50,000 impressions/month at the start. Scale to maximum reach with zero violations — in an environment where one misstep is front-page news the next morning.

The Strategy

Built a real-time content ops system: daily standups, rapid-production brief templates, a two-stage compliance gate (writer → lead → director), and a news-cycle response playbook. Every piece published before 8am. Nothing went live unsigned.

The Outcome

50K → 2.5M+ impressions in 45 days. 35,000+ new followers. 300+ pieces published, zero compliance incidents — not one. The team ran at media-company velocity inside a campaign structure. The client used this exact model as the blueprint for every campaign after.

Operations Framework

Team Build

Assembled 8-person team: writers, designers, video editors, social manager. Role clarity, daily SOP, briefing templates.

Content System

Topic clusters mapped to campaign pillars. Content calendar with 3-day buffer. Platform-native formats per channel.

Compliance Gate

Two-layer review: content lead + campaign director. Checklist-based approval. Nothing published without sign-off.

Real-Time Response

News monitoring, rapid-response templates, 2-hour turnaround for reactive content when news cycle demanded.

Daily Analytics

Morning report: reach, engagement, follower delta. Afternoon adjustment: double down on what worked yesterday.

2.5M+

Impressions · 45 Days

35K+

New Followers

300+

Content Pieces

Compliance Violations

Local SEO · Reputation Management · Multi-Location

3.8 Stars to Top-3 Google Pack — 8 Locations, 25 Staff, 4 Months

8 locations, 25+ staff trained, one reputation management system — Top-3 local pack in 4 months.

Google Business Profile Citation Cleanup Review Generation 8 Locations

3×

Visitor Growth

4.6★

Avg Star Rating

The Problem

A multi-location retail chain with 8 physical stores, average Google rating of 3.8 stars, and inconsistent online presence across directories. Competitors were outranking them in local search despite inferior products. Staff had no understanding of how their behaviour affected the store's online reputation.

The Strategy

A systematic reputation infrastructure rebuild: audit every listing across 15+ directories, unify NAP data, rebuild GBP profiles with complete information, and create a staff-level review generation process that didn't feel transactional. Then train 25+ employees to maintain it themselves.

The Outcome

3× visitor growth across all 8 locations within 4 months. Star rating from 3.8 → 4.6 with 280+ new reviews. All 8 locations reached Top-3 in the local pack for primary search terms. Results held 12+ months post-engagement — the staff maintained the system independently.

3×

Visitor Growth

280+

New Reviews

3.8→4.6

Star Rating

Top-3

All 8 Locations · Local Pack

Newsletter Growth · Email Marketing · Digital Product

I Built a Newsletter from Zero. 300+ Subscribers. 38% Open Rate. $3,800 From One Product.

Growing a newsletter from 0 to 300+ subscribers with a 38% open rate and a $3.8K product launch — all organic, all owned audience.

Beehiiv 50+ A/B Tests Zero Ad Spend 18 Months

38%

Open Rate

$3.8K

Product Revenue

The Problem

Starting a newsletter in a crowded AI space with zero audience, zero brand, and zero budget. The constraint was also the opportunity: no paid acquisition meant every subscriber had to be earned through content quality and word-of-mouth. The eventual goal was to build a monetisable audience — not just a vanity list.

The Strategy

Niche positioning (AI news for India/South Asia, not global tech noise), content-led SEO, relentless A/B testing of everything, and product validation before product creation. Build the audience first, then ask them what they need, then build exactly that and sell it back to them.

The Outcome

300+ subscribers in 18 months. Open rate of 38% — nearly double the 21% industry average — through 50+ systematic A/B tests. The Mega Prompt Bundle launched to this audience generated $3,800 in 8 months with 100+ customers across 7 countries. A self-sustaining content and revenue flywheel.

300+

Subscribers · 0 Paid

38%

Open Rate vs 21% Avg

$3.8K

Product Revenue

Countries · 100+ Customers

SaaS · Developer Tool · Side Project · 0→1

A 2am Breakdown. Six Hours Debugging a Text String. PROMPTMATRIX Was Built the Next Morning.

Turning a real engineering pain point into a SaaS product — solo, bootstrapped, building the backend in public on Instagram.

Python SDK Version Control LLM Evals 60 Days Bootstrapped

Waitlist Signups

<60d

Zero to Beta

The Problem

Every AI engineering team I spoke to had the same problem: prompts live inside application code — hardcoded strings with no version history, no testing pipeline, and no rollback capability. One bad edit breaks production. Recovery takes hours. Teams have no way to test prompt changes systematically before deploying. This is the prompt engineering equivalent of shipping code without Git.

The Strategy

Developer-first product, founder-operator GTM. Build the exact tool I wished I had, then leverage the newsletter audience for zero-cost distribution and early validation. Price for reality, not aspiration: $149 founding member (one-time, locked forever) — because engineers who cloned it on Friday are paying for the cloud by Monday. Open-source self-host free forever. Team cloud at $99/mo flat when the backend ships. Distribute through the newsletter first — 300+ subscribers who already trust the brand. Low CAC, high LTV.

The Outcome

Launched open source first — MIT licensed, browser localStorage to start, SQLite and cloud tiers coming. Building the backend in public on Instagram. Founding member pricing open — $149 one-time, locked forever. The problem is validated: anyone shipping AI products has hardcoded prompts they can't govern.

Early Access · Zero Ads

<60d

Concept to Beta

Acquisition Spend

Pricing Tiers Live

"A single bad prompt edit broke a live agent at 2am. No rollback, no history, no idea what changed. I spent 6 hours debugging a text string. That incident is why PROMPTMATRIX exists."

Jachin Saikia Sonowal

Builder, PromptMatrix · Building in public

AI Automation · SaaS Marketing · Multi-Agent Systems

Replaced a 3-Person SDR Team with 3 AI Agents. CAC Dropped 61% in 90 Days.

Multi-agent pipeline replacing a 3-person SDR team. Lead scoring → personalised outreach → adaptive follow-up. CAC cut by 61%. Sales cycle from 14 days to 5.

n8n GPT-4o HubSpot CRM Multi-Agent Zero SDRs

61%

CAC Reduction

Avg Sales Cycle

The Problem

A B2B SaaS company spending $180K/year on a 3-person SDR team generating inconsistent pipeline. Manual lead qualification meant hours of research per prospect, outreach was generic, follow-up was sporadic. CAC was rising, conversion rate was declining, and the team was burning out on low-leverage work. The ask: replace the broken system, not add more headcount.

The Architecture

Designed a three-agent pipeline with closed feedback loops: a Researcher Agent (enriches every lead with company signals, tech stack, hiring patterns, recent news), a Scorer Agent (ranks leads by ICP fit and buy-signal strength), and a Writer Agent (generates context-aware, fully personalised outreach sequences). Every reply, click, and booking feeds back into the scorer — the system learns which signals convert and reweights automatically.

The Outcome

CAC dropped 61% within 90 days. Sales cycle compressed from 14 days to 5. The pipeline running 24/7 processes 400+ leads/week — what the SDR team handled monthly. Three human roles were redeployed to closing and expansion. The system doesn't plateau: every week of data makes the scoring model sharper and the outreach more precise.

System Architecture

Signal Mapping

Mapped 14 buy-signal categories: job postings, funding events, tech stack changes, leadership transitions. Each weighted by historical close data.

Researcher Agent

n8n workflow: pulls from Apollo, LinkedIn, Clearbit, and company news APIs. Enriches each lead record with 22 structured fields in under 90 seconds.

Scorer Agent

GPT-4o prompt chain evaluates ICP match and signal heat. Outputs a 0–100 score with a 3-line rationale. Only 70+ scores enter the outreach queue.

Writer Agent

Context-aware outreach generator. References the specific signal that triggered the outreach. 5-touch sequence: email → LinkedIn → email → call brief → final email.

Feedback Loop

All reply sentiment, open rates, and bookings flow back into HubSpot and retrain the scorer weekly. System gets measurably sharper every 7 days.

61%

CAC Reduction · 90 Days

14→5d

Sales Cycle Compressed

400+

Leads Processed / Week

SDR Roles Redeployed

"We expected automation to help with volume. We didn't expect it to out-qualify our best human SDR on relevance scoring. The personalisation level was something we hadn't seen from any tool before."

Head of Revenue

B2B SaaS · Series A · 40-person team

AI Agents · Newsletter · Audience Intelligence · SaaS

24% to 52% Open Rate. One Person. Six Hours to 45 Minutes. AI Wrote the Newsletter.

Autonomous content curation, behavioural segmentation, and adaptive personalisation — 90% of editorial work handled by agents. 52% open rate. $6.2K in 90 days from a cold list.

Beehiiv Claude API Make.com Behavioural Segments Self-Improving

52%

Open Rate

$6.2K

Revenue · 90 Days

The Problem

A SaaS company running a weekly newsletter to 4,200 subscribers as a customer education channel. One person. Six hours per issue. Engagement plateauing at 24% open rate. No segmentation — everyone got the same content regardless of role, use case, or lifecycle stage. The newsletter existed but wasn't doing the job: moving subscribers toward activation, expansion, or advocacy.

The Architecture

Built a closed-loop content intelligence system with three layers: a Curator Agent (monitors 40+ sources, clusters topics by relevance signal), a Personalisation Engine (segments audience into 6 behavioural profiles based on click history and product usage), and a Synthesis Agent (writes each issue variant in brand voice, adapting angle and depth per segment). Click and reply data feeds back weekly — the system learns what content each segment actually wants.

The Outcome

Open rate from 24% → 52% in 11 weeks. Editorial time per issue: 6 hours → 45 minutes. Six personalised content variants per send, each performing 2–3× better than the previous single-version blast. $6,200 in direct newsletter-attributed revenue within 90 days from a targeted upsell sequence built into the highest-intent segment. The system gets better every week without additional effort.

Intelligence Stack

Audience Mapping

Analysed 18 months of click data to identify 6 behavioural segments: Power Users, Feature Explorers, Churners-at-Risk, Champions, Evaluators, and Passive Readers.

Curator Agent

Make.com workflow monitors 40+ RSS feeds, newsletters, and product blogs. Claude API clusters and scores articles by relevance to each audience segment daily.

Synthesis Agent

Generates 6 issue variants per week. Same core story, different angle, depth, and CTA. Power Users get technical depth. Evaluators get use-case proof. Champions get referral hooks.

Revenue Layer

Identified high-intent segment (clicked pricing content 2+ times). Built a 4-email upsell sequence delivered inside the newsletter. $6.2K attributed revenue in first 90 days.

Feedback Loop

Weekly: click patterns retrain segment profiles. Monthly: Curator Agent reweights source relevance based on what content drove the most downstream product activity.

24→52%

Open Rate · 11 Weeks

45min

Editorial Time vs 6hrs

Personalised Variants/Send

$6.2K

Newsletter Revenue · 90d

"Our newsletter went from a content obligation to a genuine revenue channel in under three months. The personalisation wasn't just better — it felt like a different product for each reader."

Head of Marketing

SaaS Platform · B2B · 4,200 Subscribers

Case Study · AI Systems Architecture · Live in Production

JACHIN

OS v8

I designed, built, and run a custom AI operating system — 25+ specialised agents across 6 departments, orchestrated as a single autonomous engine. It's running in live production right now, while you're reading this.

Not prompt wrappers. Not Zapier automations. An event-driven, memory-persistent architecture — layered security vault, self-improving feedback loops, measurable output at every stage. Weekly ops: 25 hours down to under 8.

Every agent reads and writes its own context files across sessions — a structured Markdown memory architecture I designed for marketing workloads. Each agent knows its role, its full history, and exactly what it's allowed to touch. This is what I can build for your team.

Architecture is proprietary · What you see is the shape, not the blueprint

Agents Deployed

25+

Weekly Ops Time Saved

17hrs

Email Open Rate

38%

Product Conversion Lift

+40%

Support Resolved (No Human)

90%+

Content · 7

Revenue · 4

Build · 6

Ops Automation · 3

Intelligence · 3

Security · 3

25+ AI Agents · Live Production

What you're seeing

Content ×7Writes, edits, schedules all output autonomously

Revenue ×4Runs email funnels, tracks conversions, flags drops

Build ×6Ships code, reviews PRs, monitors deploys

Ops ×3Orchestrates daily standups, task queues, SOPs

Intelligence ×3Monitors competitors, scrapes signals, briefs me

Security ×3Guards access, vaults secrets, routes all commands

Gold lines = live data packets · Hover any node for full agent brief

JACHIN

Security · Master Router

All commands route through JACHIN — access enforcement 24/7

98/100

Content

7 Agents · Daily–Weekly Triggers

RADARMonitors 30+ sources — relevance-scored digest delivered before you wake up
SCRIBEResearch-to-draft pipeline — 60% writing time reduction per newsletter
REMIXOne newsletter becomes 9 platform-native pieces — zero reformatting
HERALDPublishes to Beehiiv on schedule. Does not deviate. Does not delay.
PULSECross-platform scheduling at optimal send times per platform
GHOSTWrites SEO articles, auto-publishes, submits URLs to Search Console
CAPTUREPortable thought inbox — voice notes, screenshots, raw text → structured idea

Revenue

4 Agents · Weekly Strategic Cadence

LENSAggregates Beehiiv, Systeme.io, GSC — Monday brief with trends and anomalies
CLOSERWrites sponsorship pitches, product copy, and outreach sequences in brand voice
SCOUTScans for affiliate programs, grants, partnerships — curated list every Thursday
INTELCompetitor content gap analysis — surfaces opportunities before they trend

Build

6 Agents · Session-Triggered Pipeline

MASTER-BUILDReads system memory, identifies next build target, allocates context budget
FORGEBackend: routes, services, schemas — complete, production-ready files
PIXELFrontend: React components and responsive CSS to design system spec
PROBEAutomated test runner — hard PASS/FAIL gate. Nothing ships without this.
CLAWDebug-only — fires on PROBE failure, diagnoses root cause before re-queuing
SCRIBE-BUILDDocuments session context, updates system memory, commits to git

Ops Automation

3 Agents · On-Demand Dispatch

SCOUT-OPSSurfaces high-signal opportunities bi-weekly — pre-matched to strategic criteria
PITCHGenerates context-aware proposals and outreach sequences per target — one input, ready to deploy
COMMSDrafts responses and outbound messages in documented brand voice and communication style

Intelligence

3 Agents · Nightly / Weekly Self-Optimisation

EVOLVEReviews all agent outputs weekly — identifies underperformers, drafts improvements
ORACLEMonthly knowledge compression — archives sessions, defragments memory files
GENESISDesigns new agent roles when system gaps are detected — writes the spec

Security

3 Agents · Always On

JACHINMaster router — all commands pass through here. Enforces access by role, context, and time.
GUARDIANFilters prompt injection attempts, validates brand voice before any publish action
WATCHDOGHealth check every 6h — error alerts via Telegram, auto-triage on anomaly detection

25→8

Hours/Week Ops

25+ agents replaced 17 hours of weekly manual marketing ops. The same agent architecture is now deployable for any marketing team — content pipelines, lead scoring, CRM automation, and reporting on autopilot.

38%

Email Open Rate

AI-assisted content research, subject line testing, and send-time optimisation drove opens from 22% → 38% — nearly 2× industry average.

+40%

Conversion Lift

Lifecycle-triggered promotion agents vs broadcast. Relevance at the right moment changes everything.

90%+

Support Resolved

Customer support agent handles queries autonomously. Human escalation is the exception, not the default.

Side Project · Open Source · Prompt Management for AI Teams

Open Source · Backend in Build · Founding Access Open

PROMPT

MATRIX

v2026.1 MIT License Live

Why It Exists

2am. A single prompt edit. A live workflow down. I managed 25+ AI agents in production — no rollback, no diff, no history. Six hours to find one changed line. Three customers churned before I did. That incident is burned in. PROMPTMATRIX is built so no team ever loses that night again.

I built this for myself — 25+ AI agents, each with a hardcoded system prompt buried somewhere in a config file. Every tweak meant hunting the string, editing it, restarting the process. I pulled every agent prompt into one governed dashboard. Then open-sourced it.

Works wherever your prompts live: browser localStorage, SQLite, your own database, or the managed cloud. No vendor lock-in. If your product sends a string to a language model, it belongs in PromptMatrix.

Solo AI Builders

Browser · SQLite · Zero Setup

Open the HTML file, start adding prompts. Everything persists in localStorage. Your agents call the serve endpoint. No install, no backend, no account.

Agent System Operators

OpenClaw · LangChain · Any Orchestrator

Every agent's briefing, rules, and persona in one registry. Edit any prompt from the dashboard — no code change, no restart, no file hunting.

Teams & Startups

Govern · Version · Approve

Propose changes in plain text. Engineers review the diff and approve. Live in 10 seconds. Full audit trail. Cloud backend coming — self-host free forever.

View Live Product

promptmatrix sdk · python

$# Before PROMPTMATRIX — hardcoded, no history

prompt = "You are a support agent..."

# No rollback. No eval. No audit trail.

$# After — versioned, tested, rollback-ready

from promptmatrix import PromptClient

client = PromptClient("api-key")

prompt = client.get("support-agent")

filled = prompt.render(

context="user query", tone="helpful"

)

✓ Loaded: support-agent@v3.1.2 ✓ Eval score: 96/100 · Deployed ✓ Rollback: v3.0.8 available in 10s

BUILT FOR TEAMS WHERE AI IS THE PRODUCT

Your product makes API calls to GPT-4o, Claude, Gemini, or Llama. That means a broken prompt isn’t a bug in your backlog — it’s a churn event that already happened. PROMPTMATRIX is the version control layer that lives between your prompts and your users. For every team where prompt quality is a revenue variable, not an afterthought.

🤖

AI-Native SaaS

LLM-Powered Products

Every user interaction touches a prompt. A regression isn’t a bug — it’s a churn event that already happened. PROMPTMATRIX gives you the guardrails your deployment pipeline already has for code.

AI Support Bots Copilots RAG Systems

⚡

AI Agent Builders

Multi-Agent Pipelines

One bad system prompt in a 10-agent chain and the whole pipeline fails silently. Version every agent instruction independently. Roll back one node in seconds without touching the rest of the graph.

LangChain CrewAI AutoGen

📈

Growth & Marketing

AI Content Teams

Running AI copy, emails, and ad variants at scale means your prompts are assets — not temporary strings. PROMPTMATRIX replaces the shared Notion doc nightmare with a living, versioned prompt library your whole team can trust.

Email AI Ad Copy SEO Agents

🏗️

Series A – C Startups

Scaling AI Surface Area

Two engineers touching the same prompt with no version control is a $50K incident waiting to happen. Post-PMF teams scaling their LLM surface need the same discipline they apply to code — before the first big outage, not after.

2–20 Engineers Live LLM APIs

🛡️

Enterprise AI Teams

Governance & Compliance

Deploying AI across business units means every prompt change is a compliance event. Full audit trail, RBAC, and approval gates ensure no prompt reaches a customer-facing workflow without a sign-off chain.

RBAC Audit Trail SSO

🔬

Prompt Engineers & ML Teams

Systematic Evals at Scale

Running systematic evals across GPT-4o, Claude, and Gemini on the same prompt set is only useful if you can version and compare results. PROMPTMATRIX gives researchers a structured eval history across hundreds of iterations.

Multi-model Eval Suites A/B Testing

Prompt Version History · support-agent

Live production log — this is what PROMPTMATRIX tracks for every prompt

Commit

Change description

Eval Score

Status

a3f2c1e

Restructured resolution flow — CSAT-correlated phrasing, reduced escalation by 22%

96 / 100 ✓

Deployed

8d91b4a

Variable injection — removed 4 hardcoded strings, added product + user-tier params

92 / 100 ✓

Deployed

c7e3f0d

Extended persona block — experimental empathy tone. Eval failed threshold on edge cases.

54 / 100 ✗

Auto-Blocked

5b2a9c8

Initial version — baseline support prompt, no variable support, no tone guardrails

71 / 100 ~

Archived

Auto-eval on every push · Next run: triggered on commit

Version Control — Git for Prompts

Every commit tracked with author, timestamp, quality score, and full diff. One-click rollback in 10 seconds. The 2am debugging session that costs you customers never happens again.

# Roll back to last stable
client.rollback("support-agent", to="v3.0.8")
✓ Deployed in 10 seconds

Auto-Eval on Every Push

LLM-as-judge scoring + rule-based checks run on every commit. A prompt that scores below your threshold never reaches production. The gate holds even when humans don’t.

LLM-as-judge · Rule-based · Threshold gates

3×

3 Roles. One Platform.

Engineer, PM, Marketer — each with the right view and the right controls. No role should touch prompts they shouldn’t. No role should be blocked from prompts they need.

Prompt Studio — Test Before You Ship

Multi-model testing across GPT-4o, Claude, Gemini, Llama. Variable injection, live quality scoring, and per-call cost estimates before anything deploys. Know what it costs before it costs you.

Team Layer — RBAC · Approvals · Audit Trail

Role-based access so everyone sees exactly what they should. Approval gates before production deploys — no engineer overrides, no accidental pushes. Full audit trail answers the one question that always gets asked at 2am: who changed what and when? Cost analytics shows you spend per 300+ calls across your entire prompt fleet.

RBAC · Approval Gates · Cost Analytics · Full Audit

“One changed line. 2am. No rollback. No diff. No idea where to look. Six hours of debugging a text string while three live workflows were down and customers were churning. I built PROMPTMATRIX the next morning. No team should ever lose that night.”

Jachin Saikia Sonowal — Builder, PromptMatrix

25+ agents in production · Building in public on Instagram

Early Access Teams

Zero paid acquisition. Newsletter-first. Building in public on Instagram.

<60d

Concept to Beta

Solo. Bootstrapped. $0 spent. Frontend shipped, backend in active build.

10s

Rollback to Stable

The 6-hour debugging session becomes a 10-second command.

$149

Founding Access

One-time. Locked forever. 5 spots. Moves to $99/mo when backend ships.

Real Outcomes · Who It's Built For

Solo AI Builder

Managing 12 agent prompts across scattered config files — any change breaks something unpredictably.

PROMPTMATRIX centralises every prompt in one dashboard with full version history, serve API, and one-click rollback.

6 hrs → 10 sec to recover a broken agent

Startup AI Team

Junior dev ships a "minor tweak" to the sales email prompt — conversion drops 40% before anyone catches it.

Approval workflows block unreviewed changes from reaching production. Diffs are reviewed like PRs — in plain text, by anyone.

Zero unreviewed prompts ever reach production agents

Agencies & Operators

Running AI workflows for 5 clients — each has different prompt sets, no audit trail, no versioning, no accountability.

Multi-project registry with per-client namespacing, full change history, and automated evals that score every prompt before it ships.

Full audit trail on every prompt change, every client

Client & Professional Work

PROVEN
RESULTS

Seven engagements across D2C, AI engineering, campaign operations, digital products, and growth systems — each with a measurable outcome you can verify.

D2C Growth · 18 Months · Shopify

D2C · ASSAM TEA

Tea Brand D2C Launch

Built the entire go-to-market from zero — Shopify store, #1 keyword rankings, 5,000-subscriber email funnel. No agency. One operator. One system that kept compounding for 18 months straight.

$145K+Revenue

#112 Keywords

5KSubscribers

Read Case Study →

AI Systems · Personal OS · 25+ Agents

AI SYSTEMS · JACHIN OS

AI Agent OS — 25+ Agents Live

Built and orchestrated a personal AI operating system with 25+ specialized agents across 6 departments — Content, Revenue, Build, Intelligence, OPS, Security. Runs entirely on free-tier APIs. Designed so I approve and direct while agents execute.

25+Live Agents

6Departments

$0API Spend

Explore System →

Campaign Operations · Content Ops · Team Lead

POLITICAL · 45 DAYS

Election Campaign: 50K → 2.5M

Led an 8-person content team through a compressed 45-day election cycle. 300+ pieces produced, zero compliance violations. Scaled from 50K to 2.5M+ impressions under maximum pressure — from strategy to execution on the same day.

2.5M+Impressions

35KNew Followers

0Violations

Read Case Study →

Newsletter Growth · Email Marketing · Digital Product

NEWSLETTER · BEEHIIV

Matrix Labs — 0 to Revenue

Grew a newsletter from 0 to 300+ subscribers with 38% open rate — nearly double the industry average — through 50+ systematic A/B tests. Product launch generated $3.8K across 7 countries from a ₹5,000 Meta ad spend.

38%Open Rate

$3.8KProduct Revenue

7Countries

Read Case Study →

SaaS Side Project · Developer Tool · 0→1

SAAS · LIVE BETA

PROMPTMATRIX — From Pain to Product

Turned a real engineering pain point into an open-source tool — solo, bootstrapped, building in public. Prompt governance and version control for AI products. Browser, SQLite, or cloud — pick your storage. Founding member pricing open now.

5+Waitlist Teams

<60d0 → Beta

8Countries

Try PROMPTMATRIX →

Local SEO · Reputation Management · Multi-Location

LOCAL SEO · 8 LOCATIONS

Multi-Location Retail: Top-3 in 4 Months

Rebuilt the reputation infrastructure for 8 retail locations. Star rating from 3.8 → 4.6 with 280+ new reviews. All 8 locations Top-3 in local pack. 3× visitor growth. Sustained 12+ months with no ongoing spend.

3×Visitor Growth

4.6★Avg Rating

Top-3All 8 Locations

Read Case Study →

Digital Product · Newsletter Drop · Paid Amplification

DIGITAL PRODUCT · SYSTEME.IO

AI Prompt Bundle — matrixlabs.systeme.io

Designed and launched the Mega Prompt Bundle — distributed through Matrix Labs newsletter (300+ subscribers, 38% open rate) and a ₹5,000 Instagram Meta ad campaign. Built the complete product + distribution stack from scratch. Buy at matrixlabs.systeme.io.

3.8×ROAS

₹5KAd Budget

2Channels

Get the Bundle →

1 / 7

Growth System Thinking

EXPERIMENT
LOG

Real experiments. Real outcomes. The ones that worked and the ones that didn't — because the thinking is visible in both.

WIN RATE

70%

Showing: All Experiments 10 experiments

nothing matching that filter — yet

EMAIL 2024 ↑ +81% OPEN RATE

Subject Line Personalisation at Scale

Hypothesis Subject lines with a first-name variable + a number will outperform generic curiosity lines for an AI audience.

Result 22% → 38% open rate over 50+ tests. Personalised + numeric subject lines won 62% of head-to-heads. Generic curiosity lines collapsed after list >500.

✓ Shipped to permanent system

GROWTH 2021 ✕ FAILED → PIVOTED

Cold Outreach at Scale vs Warm Referral Engine

Hypothesis A high-volume cold outreach system (200+ DMs/week via LinkedIn + WhatsApp) will consistently fill the client pipeline faster than relying on referrals from existing clients.

Result Cold volume produced leads, but close rate collapsed to 4% and client quality was poor — high churn, scope creep, price sensitivity. Meanwhile, a single referral chain from 3 existing clients brought in 11 projects at 3× the LTV. The infrastructure investment in cold outreach was real but the signal was wrong: volume ≠ pipeline quality. Rebuilt the system around referral triggers — milestone emails, outcome reports shared publicly, a formal "refer a contact" ask at project close. Referrals hit 68% of new business within 6 months.

✗ Cold volume killed. Referral engine became the primary acquisition system ever since.

SEO 2022 ↑ #6 → #1 in 4 months

Topic Cluster Architecture vs Isolated Posts

Hypothesis A pillar + cluster architecture will outrank isolated well-optimised posts for long-tail D2C keywords.

Result Ranked 12 high-intent keywords in top 6 within 4 months. Organic traffic 3×. The cluster built internal link equity that individual posts never achieved alone.

✓ Core methodology ever since

PRODUCT 2024 ↑ $29 beat $19 and $49

Pricing Anchor Test — Mega Prompt Bundle

Hypothesis $29 will have better volume × margin than $19 (too cheap = low quality signal) and $49 (conversion drop).

Result $29 tier generated 2.3× the revenue of the $19 anchor and 1.8× the $49 tier. $29 hit the psychological "fair trade" window for information products in this audience.

✓ $29 set as permanent price

AI 2025 ↑ 90% query deflection

AI Support Agent on Digital Product Store

Hypothesis An AI agent with product FAQ context can handle most support queries without human input, scaling the product without scaling support time.

Result 90%+ of incoming queries resolved without escalation. Manual support time dropped from ~2h/week to ~12min/week. Satisfaction maintained at 4.8★.

✓ Running live

GROWTH 2024 ✕ FAILED

Landing Page Conversion — Long-Form vs Short

Hypothesis A longer, more detailed landing page (social proof, FAQ, risk reversal) will convert better than a minimal 3-section page for cold newsletter traffic.

Result Short page won: 8.5% CVR vs 6.1% for the long version. Cold traffic doesn't read — it scans for the single signal it needs. Long-form works post-trust, not pre-trust.

✗ Rolled back. Short page permanent.

EMAIL 2024 ↑ 18% subscriber recovery

Re-Engagement Sequence for Cold Subscribers

Hypothesis A 3-email re-engagement sequence with a clear "stay or go" CTA will recover or cleanly remove cold subscribers, improving list health.

Result 18% of 30-day inactive subscribers re-engaged. Clean unsubscribes dropped the list by ~6% but open rate increased from 31% → 38% post-cleanup.

✓ Quarterly system now

AI 2025 ↑ 60% writing time saved

AI-Assisted Newsletter Pipeline (SCRIBE Agent)

Hypothesis An AI agent that consumes curated research and outputs a structured draft in my voice will cut production time without degrading quality or reader trust.

Result 60% reduction in writing time. Open rate maintained at 38%. 3 readers specifically mentioned "the writing feels more consistent" — the agent enforced my style guide more reliably than I did manually.

✓ SCRIBE agent permanent

SEO 2023 ✕ FAILED

Programmatic Landing Pages for Local SEO

Hypothesis Auto-generated location × service landing pages will capture long-tail local queries at scale without proportional content effort.

Result Google deindexed 40% of the pages within 6 weeks. Thin content penalty. Manual high-quality pages outperformed at 1/10th the scale. Programmatic only works with genuine content differentiation per page.

✗ Scrapped. Lesson: quality signals beat volume at scale.

GROWTH 2023 ↑ +60% BOOKINGS

Paid + Organic Combined vs Each Channel Solo — Healthcare Clinic

Hypothesis Running Google Ads and local SEO simultaneously for a healthcare clinic will compound results beyond either channel's individual ceiling — ads capture intent while SEO builds trust signal that lowers CPC over time.

Result Appointment bookings up 60% in 4 months. GBP star rating held at 4.8 throughout. The combined stack worked exactly as theorised — organic trust reduced bounce on paid landing pages, dropping CPA by 34%. Either channel alone had plateaued at +18%.

✓ Dual-channel model now standard for service clients

10 Experiments

7 Shipped

3 Killed

70% Ship Rate

Every failed experiment is documented — the thinking is the asset, not just the win.

AI & Agents	OpenClaw, ChatGPT API (GPT-4o), Claude API, Perplexity AI, Custom GPTs, PromptMatrix (Builder), Advanced Prompt Engineering Frameworks
Dev & SDK	Python, Python SDK development, REST API design, GitHub Actions, Ghost Admin API, Beehiiv API, Gumroad API, GA4 Reporting API, Anthropic API, OpenAI API
Automation	n8n (self-hosted), Make.com, Zapier, API webhooks, cron-triggered pipelines, JSON orchestration
Analytics & SEO	Google Analytics 4 (Certified), Search Console, GTM, Ahrefs, SEMrush, Screaming Frog, Hotjar, Looker Studio
Email & CRM	Beehiiv, ActiveCampaign, Klaviyo, Mailchimp, ConvertKit, HubSpot (Certified), Brevo, Salesforce basics
E-Commerce & Web	Shopify, WooCommerce, WordPress, Webflow, Ghost, GitHub Pages, HTML/CSS, Schema Markup
Design & Content	Canva, Figma, Adobe Photoshop, Illustrator, CapCut, Descript, OBS Studio, StreamYard, Loom

JACHIN
SAIKIA
SONOWAL

The problem I
reliably eliminate.

WHAT I'M
NOT

THE
JOURNEY

HOW
I WORK

CASE
STUDIES

PROVEN
RESULTS

SKILLS &
TOOLS

TOOLS
I USE

RESULTS,
IN THEIR WORDS

CERTI
FICATIONS

ACADEMIC
BACKGROUND

LET'S BUILD
GROWTH
SYSTEMS

EXPERIMENT
LOG

JACHIN SAIKIA SONOWAL

The problem Ireliably eliminate.

WHAT I'MNOT

THEJOURNEY

HOWI WORK

CASESTUDIES

PROVENRESULTS

SKILLS &TOOLS

TOOLSI USE

RESULTS,IN THEIR WORDS

CERTIFICATIONS

ACADEMICBACKGROUND

LET'S BUILDGROWTHSYSTEMS

EXPERIMENTLOG

JACHIN
SAIKIA
SONOWAL

The problem I
reliably eliminate.

WHAT I'M
NOT

THE
JOURNEY

HOW
I WORK

CASE
STUDIES

PROVEN
RESULTS

SKILLS &
TOOLS

TOOLS
I USE

RESULTS,
IN THEIR WORDS

CERTI
FICATIONS

ACADEMIC
BACKGROUND

LET'S BUILD
GROWTH
SYSTEMS

EXPERIMENT
LOG