AI-Powered Sales & Operations Agent

Building a fully autonomous AI sales agent integrated with Slack, HubSpot CRM, vector search, and cold calling capabilities to automate end-to-end business development operations.

AI-Powered Sales & Operations Agent

Client Overview

About the Project

A B2B services company was spending significant hours each week on repetitive sales tasks: manually researching leads, updating CRM records, scheduling outreach sequences, and compiling sales collateral. The sales team was reactive rather than strategic, constantly caught in administrative coordination instead of high-value conversations. Their core challenge was fragmentation. Lead data lived in disconnected spreadsheets, HubSpot records were updated inconsistently, cold outreach was handled manually one contact at a time, and internal knowledge for crafting pitches was scattered across documents with no fast retrieval method. This prevented the team from operating at the scale the business required, and the cost of hiring additional SDRs to handle volume was prohibitive.

Our Approach

The Solution

Zentric Solutions designed and built Elliot Smith — a persistent AI sales and operations agent running on n8n, triggered through the team's existing Slack workspace. Rather than building a standalone tool, the agent was embedded directly into daily communication workflows, allowing the team to assign tasks conversationally and receive structured outputs without leaving Slack. The agent was connected to a custom AI leads database with natural language query capability, enabling Elliot to research, filter, and surface qualified prospects on demand. A Supabase vector store was built and loaded with company documents, case studies, and pitch materials, giving the agent semantic search over institutional knowledge when crafting personalised outreach. HubSpot integration allowed Elliot to create and update contact records, manage list memberships, and retrieve contact details in real time during conversations. A dedicated Blog Post Agent subagent enabled automatic content generation on request, while a Cold Calling Agent subagent handled scripted outreach workflows for targeted prospect lists. PostgreSQL-backed chat memory ensured continuity across sessions, allowing Elliot to remember context, prior instructions, and ongoing tasks without repetition from the operator.

Tech Stack

n8nOpenAI GPT-4Slack APIHubSpot CRM APISupabase Vector StorePostgreSQLREST APIsWebhook ServicesCloud Infrastructure

Have a similar idea?

We turn ambitious products into reality. Let's talk about yours.

Get in Touch

Project Tags

n8nAI AgentHubSpotSlackAutomationSalesVector SearchOpenAI

Portfolio

More Case Studies

Common Questions

Frequently Asked Questions

Everything you need to know about this project and our approach.

Elliot is triggered directly through Slack. Team members message the bot in natural language — requesting lead research, contact creation, list updates, or content drafts — and the agent handles the task autonomously, returning structured outputs in the same channel.

A custom AI leads database was built and connected to the agent with natural language query support. Elliot can search, filter, and retrieve leads based on criteria specified in the Slack message, without requiring manual database access.

Yes. A dedicated Blog Post Agent subagent handles content generation on request, drawing on the vector store of company materials for accurate, on-brand output. Cold outreach emails are similarly generated using retrieved lead context and stored pitch frameworks.

Yes. PostgreSQL-backed persistent chat memory allows Elliot to maintain context across sessions. Ongoing tasks, prior instructions, and conversation history are preserved, so the team does not need to repeat context each time they engage the agent.

Elliot can search existing contacts, create new records, update contact properties, add or remove contacts from lists, and retrieve contact details by ID — all triggered through a natural language instruction in Slack with no manual CRM navigation required.

Yes. The agent operates with least-privilege API access, all credentials are stored in environment variables outside the workflow, and Slack message authentication is validated before any action is taken. Access is scoped to the specific HubSpot objects and database tables required.

Smart IT Solutions for Modern Businesses

Zentric Solutions delivers cutting-edge digital products that streamline operations, enhance engagement, and drive lasting growth.

Let's Collaborate