💬 📅
AI Automation Architect · Delhi, India · Available Globally

Ashutosh
Sharma
AI Automation.

I design and build production-grade AI automation pipelines — B2B sales intelligence, agentic workflows, self-healing systems, document processing and more — on Make.com and connected platforms. 25 years of enterprise operations experience means I build for what fails in production, not just what works in a demo.

28+ Scenarios Built · 8+ Enterprise Clients · 25 Years Ops Background · Global Delivery
By the numbers
0
Make.com scenarios built across industries
0
Enterprise clients — Hexaware, Kotak, Arvind Group
0
Years in enterprise operations — BT, Telstra, SITA
0
Agentic AI architectures — static, evolving, self-healing
// Typical pipeline architecture
Signal / Webhook TriggerINPUT
Normalise + Schema MapTRANSFORM
Lookup / Duplicate CheckVALIDATE
API EnrichmentENRICH
AI Classification + RoutingDECIDE
Action + Delivery + LoggingEXECUTE
Delivered for Hexaware Technologies Kotak Mahindra Bank Arvind Group MSME — Govt of India iOPEX Technologies / BT
01 · Background & Expertise

25 years of operations.
Applied to the AI era.

Not a theoretical AI consultant. An operations leader who spent two decades building and running complex systems for BT, Telstra, SITA, and Nortel — and now applies that same thinking to designing AI automation that works in the real world.

🌐
Enterprise Operations — 25 Years

Senior technical operations across BT, Telstra, SITA, and Nortel spanning telecoms, IT services, and digital transformation. Led CXM platform migrations (Genesys Engage to Cloud), built US telco offshoring frameworks, and managed long-running enterprise delivery partnerships. Currently Senior Technical Analyst at iOPEX Technologies managing the BT relationship.

BT TelecomTelstraSITANorteliOPEXCXMGenesys
🤖
Enterprise AI Training & Consulting

Founder of Optivantage Technologies. Designs and delivers Gen AI and Agentic AI training programs for enterprise clients — Hexaware, Arvind Group, Kotak Mahindra Bank, MSME Government of India. Builds custom AI awareness curricula, prompt engineering frameworks, and live demonstration scenarios adapted to each organisation's context and maturity level.

Gen AI TrainingAgentic AIPrompt EngineeringEnterprise ConsultingOptivantage
⚙️
AI Automation Architecture

28+ production Make.com scenarios spanning B2B sales pipelines, support triage, document intelligence, HR automation, voice AI, self-healing and self-evolving agents. Builds automation that handles real complexity — multi-step API chains, conditional routing, error recovery, and retry logic. Documentation and handover included on every engagement.

Make.comOpenAIClaudeAgentic SystemsSelf-HealingSelf-Evolving
Platforms & Tools
⚙️
Make.com
Orchestration
🧠
OpenAI / ChatGPT
AI Generation
🤖
Anthropic Claude
AI Reasoning
🎯
HubSpot
CRM Automation
🔍
Clay / Apollo
Enrichment
ZeroBounce / NeverBounce
Email Validation
📧
Instantly / Lemlist
Outreach Delivery
🔔
Slack / Gmail / Outlook
Notifications
📁
Google Drive / OneDrive
Document Processing
🎙️
Bolna AI
Voice AI / KYC
📊
Google Sheets / Excel
Data Layer
🔗
HTTP / REST APIs
Custom Integrations
🌐
Perplexity AI
Real-time Research
🔎
Leadfeeder / RB2B
Visitor Intelligence
📋
PDF.co
Document Extraction
02 · Use Cases by Demand Horizon

Not just what's trending
what businesses keep paying for.

Every NOW use case is already built and deliverable. Every NEXT use case is a direct extension of existing work. FUTURE use cases reflect where I am actively building skills and positioning — so when that demand arrives, I'm already there.

🎯✓ Built
B2B Sales Intelligence Pipeline

Signal-to-outreach automation. Deanonymise site visitors, pull intent signals, enrich via API, validate email through 5-branch logic, create CRM contact, deliver personalised outreach sequence. End-to-end with zero SDR involvement.

Make.comHubSpotClay/ApolloZeroBounceLemlist/Instantly
→ Full blueprint built. 16-module scaffold ready to import.
✍️✓ Built
AI SDR Email Coach

Webhook receives lead context. Pulls versioned prompt templates from a live spreadsheet. A/B routes between prompt variants. Generates personalised first-touch cold emails per lead. Marketing teams update messaging without touching the code.

Make.comWebhookDynamic PromptsA/B RoutingOpenAI
→ Live scenario with versioned prompt architecture in portfolio.
🎫✓ Built
IT & Support Ticket AI Triage

Webhook receives ticket. AI classifies type, priority, SLA risk. Routes to correct Slack channel. Escalates critical tickets automatically. Built and live for Hexaware — 9 executions, 28 operations. Adaptable to any service desk setup.

Make.comOpenAISlackWebhookAI Classification
→ Live. Hexaware client. 9 executions confirmed.
📄✓ Built
Document Intelligence Pipeline

Triggers on new PDF or DOCX upload. Extracts text, AI analyses and scores, routes output to CRM, Slack or Sheets. Built for SDLC analysis and financial results. The same pattern works for contracts, invoices, proposals, and compliance documents.

Make.comPDF.coGoogle DriveOneDriveAI Agent
→ Two variants: SDLC Analyser + Financial Results Analyser (Microsoft).
🧑‍💼✓ Built
AI Recruiting & HR Pipeline

Monitors drive folder for new CVs. Extracts content. AI scores against job criteria. Strong match triggers recruiter email; weak match logs to review sheet. Eliminates manual first-pass screening for hiring teams at any scale.

Make.comGoogle DrivePDF.coAI AgentGmail
→ Live scenario with full conditional routing on match quality.
📊✓ Built
Meeting Intelligence & Action Tracking

Monitors cloud storage for new notes or recordings. AI extracts action items, decisions, and owners. Sends formatted summary by email. Logs to sheets. Saves 20–40 minutes of manual write-up per meeting across the whole organisation.

Make.comOpenAIGoogle DocsOneDriveGmail/Outlook
→ Two variants built: Google Workspace + Microsoft 365.
🔁✓ Built
Self-Healing Agentic Pipelines

Pipelines that detect their own failures and recover. AI classifies the error, selects a recovery path, retries with corrected parameters, escalates to human fallback only if retry fails. Production automation that doesn't need babysitting.

Make.comOpenAIError ClassificationRetry Logic
→ Live: Simple Demo + Fixed production variant both built.
🧬✓ Built
Self-Evolving AI Agent

Agent rewrites its own system prompt based on outcome feedback. Bad runs trigger AI-driven improvement. Good runs lock the version. Gets smarter every cycle without manual intervention — a foundation for autonomous AI improvement at scale.

Make.comClaude/OpenAIGoogle SheetsFeedback Loop
→ Live: DEMO 3 + Agent 3a Evolution Engine (production variant).
🎙️◎ Ready to Build
Voice AI & KYC Automation

AI conducts customer verification or onboarding calls. Webhook processes logs — AI extracts structured data, flags anomalies, routes escalations. Compliance-ready audit trail auto-generated. High demand in BFSI and customer operations.

Bolna AIMake.comOpenAIWebhook
→ Bolna KYC Call Logs scenario already in portfolio.
📡◎ Ready to Build
Proactive Business Intelligence

Instead of waiting for questions, the agent monitors data sources on a schedule, detects anomalies or threshold breaches, and pushes insights to Slack before anyone notices. Reactive reporting becomes proactive operations intelligence.

Make.comPerplexity AISlackScheduled Triggers
→ Extends newsletter automation + sentiment analyser patterns.
📬✓ Built
Email Intelligence & Auto-Response

Monitor inbox. Classify by intent and urgency. Route to the right team or draft a response automatically. Log sentiment trends. Escalate anomalies. Reduces email handling time by 60–80% for high-volume operations inboxes.

Make.comGmail/OutlookOpenAISentiment Analysis
→ Live: Email Response Agent (Self-Evolving) + Sentiment Analyser (Gmail + Outlook).
🏗️◎ Ready to Build
Multi-Agent Task Orchestration

A coordinator agent delegates complex tasks to specialist sub-agents — research, writing, fact-checking, formatting. Each runs independently, returns results, the coordinator synthesises. Humans only see the final output. Scales output without scaling headcount.

Make.comWebhook ChainsClaudeOpenAI
→ Foundation: self-evolving + self-healing architectures already delivered.
🏢◎ Positioning Now
AI-Native Operations Centers

NOC/SOC replacement for L1 operations. Incidents detected, classified, escalated, and partially resolved by AI before a human is paged. Runbook-driven autonomous remediation. The logical evolution of triage + self-healing work already delivered.

Multi-AgentIncident MgmtRunbook AIReal-time Alerts
→ Direct extension of Hexaware triage + self-healing agent. Telecom ops background applies.
⚖️◎ Positioning Now
Real-Time Compliance Monitoring

AI monitors transactions and communications against policy rules in real time. Flags anomalies before violations occur. Auto-generates audit trails. Critical for BFSI — natural extension of the SDLC document analyser already built for banking.

Policy EngineReal-time MonitorAudit TrailBFSI
→ SDLC analyser for banking already delivered at Hexaware.
🧠◎ Positioning Now
Company Knowledge Agents

An AI that knows your company's policies, processes, and history — answers employee questions in real time. Auto-updates when documents change. Onboards new hires. Reduces L1 HR and IT support queries by 70%+. The internal helpdesk of the next decade.

RAG ArchitectureKnowledge BaseClaude/OpenAISlack/Teams
→ Foundation: RAG and NotebookLM in enterprise AI training curriculum.
🔄◎ Positioning Now
Autonomous Revenue Operations

CRM stays clean without human effort. Deals auto-progress on signal triggers. Pipeline gaps flagged proactively. Forecast anomalies raised before the board meeting. RevOps teams shrink while output grows — the full automation of the GTM function.

CRM AutomationHubSpotPipeline AISignal Triggers
→ Direct extension of sales pipeline + SDR automation in portfolio.
🌐◎ Positioning Now
Omnichannel Journey Orchestration

A single signal — visit, email open, support ticket, or social mention — triggers a coordinated response across email, WhatsApp, and Slack. The experience feels personal and timely. Every touchpoint logged, measurable, and optimised continuously.

Multi-channelWhatsApp APIEmail SequencesCRM Sync
→ Extension of signal pipeline + email automation patterns proven in portfolio.
🔧◎ Positioning Now
AI Procurement Intelligence

Monitor supplier signals, price changes, and availability automatically. AI generates RFQs, evaluates responses, flags anomalies. Connects to ERP via API. Reduces procurement cycle time significantly — directly relevant to distribution and manufacturing verticals.

Supply Chain AIAPI IntegrationSupplier MonitorERP Connectors
→ Background in B2B distribution (Vishal Enterprises) and electronics procurement.
03 · Market Demand Overview

Where the market is going
and where I already am.

NowActive, paying demand
B2B sales intelligence & outbound
Support ticket AI triage
Document processing & extraction
HR and recruitment screening
Meeting intelligence & summaries
Email sentiment & auto-response
NextProcurement happening today
Self-healing production pipelines
Voice AI and KYC automation
Proactive BI and alerting agents
Self-evolving prompt systems
Multi-agent task orchestration
Real-time compliance signals
FuturePositioning and skilling now
AI-native operations centers
Autonomous revenue operations
Company knowledge agents
AI procurement intelligence
Omnichannel journey orchestration
Autonomous business monitoring
04 · Proof of Work — Real Scenarios

28 live scenarios. A selection of
the most representative builds.

Scenario Stack Relevance What it demonstrates
SDR Email Coach
Webhook → Versioned Prompts → A/B Router → Email
Make.comWebhookGoogle SheetsOpenAI
Sales outbound automation
Prompt versioning without redeploy. A/B routing on live data. Personalised outreach at scale without manual SDR involvement.
Hexaware — AI Support Triage
Webhook → AI Classify → Route → Slack Alert
Make.comOpenAISlack
Signal routing & triage
Live client build. 9 executions. AI classification with conditional routing and human escalation path already in production.
AI Recruiting Pipeline
Drive → PDF Extract → AI Score → Route → Log
Make.comPDF.coHTTP APIAI Agent
Multi-step pipeline architecture
Signal → enrich → validate → route → act. The fundamental pattern behind any complex automation pipeline.
Self-Healing Agent
Process → Detect Failure → Classify → Retry → Escalate
Make.comOpenAIJSON
Agentic error recovery
Pipelines that recover from failure autonomously. Retry logic, fallback routing, and human escalation built in.
Self-Evolving Agent (Dynamic Prompt)
Feedback Loop → AI Rewrites Prompt → Version Lock
Make.comOpenAIGoogle Sheets
Autonomous AI improvement
Agents that improve themselves over time. Rare capability that demonstrates next-generation automation thinking.
Bolna Voice AI — KYC Call Logs
Call Webhook → AI Extraction → Structured Log → Sheets
Bolna AIMake.comOpenAI
Voice AI pipeline
Voice AI integration with structured data extraction and compliance logging. Growing demand especially in BFSI and customer ops.
05 · Frequently Asked Questions

Questions before the call.

What kind of AI automation do you build? +

Production-grade Make.com pipelines — B2B sales intelligence, support triage, document processing, HR screening, self-healing agents, voice AI integrations, and agentic systems with feedback loops. Every build includes full error handling, documentation, and handover notes.

Are you available for freelance and international projects? +

Yes. Based in Dwarka, New Delhi, India. Delivering globally — clients across India, UK, USA, Southeast Asia, and the Middle East. All project work is fully remote with clear async communication. Book a discovery call to discuss your project and timeline.

What platforms and tools do you work with? +

Make.com as the primary orchestration layer. Integrations include OpenAI, Anthropic Claude, Google Workspace, Microsoft 365, HubSpot, Apollo, Clay, ZeroBounce, NeverBounce, Instantly, Lemlist, Slack, Bolna AI, Perplexity AI, and custom HTTP API connections. I adapt to the client's existing stack where possible.

How long does a typical automation project take? +

Simple automations (3–5 modules, single workflow) can be delivered in days. Complex pipelines — multi-step API chains, conditional routing, error recovery, and CRM integration — typically take 2 to 4 weeks to scope, build, test, and hand over. Book a call for a scoped estimate specific to your requirements.

What is the difference between Make.com automation and agentic AI? +

Make.com automation follows pre-defined rules and conditional logic. Agentic AI goes further — the AI makes contextual decisions, detects its own failures, rewrites its own prompts, and takes autonomous action without a human intervening. I build both, and have delivered self-healing and self-evolving agents that combine both approaches.

Do you provide documentation and handover support? +

Yes, on every engagement. Each scenario ships with inline module documentation, clearly labelled API swap points, and handover notes your team can follow. Training sessions are available as an add-on. The goal is always that your team can maintain and extend the build independently after delivery.

Let's talk about
your automation
requirements.

Whether you have a clear brief or just an idea of the problem you want to solve — book a 30-minute discovery call. No obligation, no pitch deck. Just a direct conversation about what you need and whether I'm the right person to build it.

📅 Book a 30-Min Call 💬 WhatsApp Me ✉️ Email Instead