EnsightLabs

AI that answers, creates, converts, and grows your business.

Comment

Inspiration

EnsightLabs was inspired by a problem we repeatedly saw among growing businesses and building a personal brand as a coach. Opportunities were being lost because customers could not get accurate answers quickly enough probably because the customer is reaching out during odd houyrs.

A potential customer might visit a website after business hours with questions about pricing, policies, product suitability, or the next step. If no one responds, that visitor often leaves. Traditional chatbots do not solve this problem because they rely on rigid scripts and rarely understand the business. Generic AI tools can respond naturally, but they may provide inaccurate information because they are not grounded in the company’s actual knowledge.

We wanted to build something more useful, an AI-powered growth platform that understands each business, communicates with its customers, performs approved actions, creates content, and helps the business grow without requiring a large team which often times can incure extra cost for growing busineses.

That vision became EnsightLabs.

What it does

EnsightLabs gives businesses AI tools that support customer engagement, sales, marketing, and growth.

A business can create a custom AI website agent, choose between chat and voice interactions, customise its appearance, and train it using approved documents, policies, pricing information, FAQs informations.

Once deployed, the AI agent can:

Answer customer questions using the business’s private knowledge base.

Communicate through natural chat or voice conversations.

Capture and qualify potential customers and notify the business owner through email.

Collect structured information such as names, email addresses, needs, and budgets.

Score and prioritise leads based on their level of interest.

Check calendar availability and book meetings.

Create Google Meet links and send calendar invitations.

Preserve conversation histories for review and improvement.

Help the business draft relevant marketing and social media content.

EnsightLabs is more than a chatbot. The goal is not simply to generate responses, but to help businesses move customers from curiosity to action.

For example, when a visitor asks about a service, the agent can answer from the company’s documents, ask relevant follow-up questions, identify the visitor as a qualified lead, send emails to business owner and also help them book a meeting all within one conversation.

How we built it

EnsightLabs was built as an AI-powered business operations platform for small and growing businesses. Instead of functioning as a conventional chatbot, it acts as a digital front office: answering customer questions, identifying sales opportunities, scheduling meetings, producing marketing content, and helping businesses assess their financing readiness.

From a social-science perspective, EnsightLabs addresses an organizational problem. Many small businesses have valuable knowledge about their products and customers, but that knowledge is often scattered across documents or held by a few employees. EnsightLabs converts this knowledge into a shared digital resource that can be used consistently across customer service, sales, marketing, and business planning.

How the Project Was Built

Businesses create an AI agent and upload PDF, DOCX, or TXT documents containing information such as product descriptions, prices, policies, and frequently asked questions. These documents are divided into smaller sections and transformed into vector embeddings.

When a visitor asks a question, EnsightLabs uses Retrieval-Augmented Generation (RAG) to find the most relevant information from that business’s private knowledge base. The retrieved information is then provided to Gemini so the response remains grounded in the business’s actual documents. Each business and agent has an isolated knowledge base, reducing the risk of information from one organisation appearing in another organisation’s responses.

The platform supports both text and voice conversations. It can automatically detect a visitor’s language, maintain conversational history, stream responses in real time, and provide spoken answers. Conversations are stored and later analysed to identify customer intent, qualify leads, and produce business analytics.

Technologies Used

The frontend was developed with: Next.js 16 React 19 TypeScript Tailwind CSS Supabase JavaScript client

The backend was developed with: Python 3.12 FastAPI WebSockets for real-time chat and voice streaming LangChain for AI workflow orchestration Chroma as the persistent vector database Supabase Authentication and PostgreSQL Docker for containerisation

GitHub Actions and GitHub Container Registry for deployment workflows Additional business integrations include Paystack for subscriptions and payments, ZeptoMail for transactional emails, and Discord webhooks for operational alerts.

Google Cloud Products and Technologies

Google technologies form the core intelligence and communication layer of EnsightLabs: Vertex AI: Provides the managed environment through which the application accesses its generative AI and embedding models.

Gemini 2.5 Flash on Vertex AI: Generates grounded customer responses, extracts structured lead information, produces marketing content, and creates financing-readiness assessments.

Vertex AI text embedding model (text-embedding-004): Converts business documents into numerical representations that can be searched semantically during Retrieval-Augmented Generation.

Google Cloud Speech-to-Text: Transcribes a visitor’s spoken question and supports language detection for voice conversations.

Google Cloud Text-to-Speech: Converts the AI-generated response into natural MP3 audio for the visitor.

Google Calendar API: Reads a business owner’s availability and creates calendar events when a visitor requests a meeting.

Google Meet: Meeting links are generated through the Google Calendar API and added to invitations automatically.

Google OAuth 2.0: Allows business owners to securely connect their Google accounts and grant limited Calendar permissions.

Google Cloud Application Default Credentials and Identity and Access Management (IAM): Secure the backend’s access to Vertex AI and the speech services.

How AI-Native Operations Affect the Business

EnsightLabs makes AI part of the business workflow rather than treating it as a separate assistant. A customer conversation can become several operational outcomes: an answered question, a recorded demand signal, a qualified lead, a meeting, or an insight for future marketing. This changes the organisation in several ways: Redistribution of labour: Routine enquiries and initial sales qualification are automated, allowing employees to focus on negotiations, complex support, and relationship-building.

Continuous availability: Small businesses can respond across time zones and languages without maintaining a large support team.

Institutional memory: Business knowledge is moved from individual employees into a reusable, searchable system, making customer communication more consistent.

Data-informed decisions: Conversations reveal what customers want, where they are confused, and which prospects show purchasing intent.

Lower barriers to participation: Multilingual text and voice interfaces make the business more accessible to people with different languages, literacy preferences, or interaction habits.

Alternative evidence for financing: Customer conversations, qualified leads, booked meetings, and platform activity can be organised into signals of demand and business traction. These signals support financing readiness, but they are not treated as proof of revenue or a guarantee of credit.

Importantly, the system maintains human oversight. Contact details are retained only when supported by the conversation, low-confidence leads are flagged for review, generated content requires approval, and financing assessments are presented as decision support not lending decisions.

Challenges we ran into

One of our biggest challenges was ensuring that the AI gave accurate answers based on each business’s approved information and also reduce response time when a user opts for using the voice agent. A response can sound convincing while still being wrong, so we had to build a reliable retrieval process that searches the business’s private knowledge base and provides the most relevant context before generating an answer.

Performance was another challenge. The agent needs to retrieve information, understand the visitor’s intent, generate a response, and sometimes perform an action all within a few seconds. We improved this by reducing unnecessary context and LLms round trips, running independent operations concurrently, optimising database queries, and moving longer tasks to background jobs.

Voice conversations introduced additional complexity. Speech recognition, AI response generation, and text-to-speech must work together quickly for the interaction to feel natural. We also had to account for different accents, background noise, pauses, and interrupted speech.

Marketing has also been a major challenge. As a small, self-funded founder, I do not yet have the budget required to run large advertising campaigns or compete aggressively for attention. This has made it difficult to reach as many businesses as we believe could benefit from EnsightLabs. I have had to rely on organic content, direct outreach, product demonstrations, customer conversations, and referrals. Although this has limited the speed of our growth, it has helped us build closer relationships with early potential customers.

Accomplishments that we're proud of

Seeing EnsightLabs grow from an idea into a live, working product is one of our proudest accomplishments. It is exciting to know that businesses can now create an AI agent, upload their business knowledge, and use the platform to engage customers and automate important tasks.

Securing our first paying business customer was another major milestone. It showed us that EnsightLabs is not only an interesting idea but a product that a real business finds valuable enough to pay for and use. We are now focused on getting more people to test the platform, review it, and share their honest thoughts so we can continue improving it.

We are also proud of what we learned while building with Google Cloud technologies. We used Vertex AI and Gemini 2.5 Flash to generate grounded responses, extract structured lead information, create marketing content, and produce financing-readiness assessments. We used the text-embedding-004 model to build a Retrieval-Augmented Generation system that can search a business’s documents semantically before answering customer questions.

Building the voice experience taught us how to combine Google Cloud Speech-to-Text, language detection, Gemini, and Google Cloud Text-to-Speech to create natural voice conversations. Integrating Google OAuth 2.0, Google Calendar, and Google Meet taught us how to manage secure permissions, check real-time availability, create calendar events, and generate meeting links automatically.

Building and launching EnsightLabs on a limited budget has shown us that progress is not measured only by large user numbers. Taking an idea, solving difficult technical problems, securing a paying customer, and turning everything into a product that businesses can use are meaningful achievements. Our next goal is to earn the trust of more businesses, personal brands, coaches and build a community of customers who use, recommend, and help us improve EnsightLabs.

What we learned

We learned that many small businesses and brand coaches do not struggle because they lack knowledge or valuable services. They struggle because they have limited time, small teams, and too many responsibilities competing for their attention.

A brand coach may be creating content, responding to enquiries, following up with leads, managing clients, and booking consultation calls alone. Small-business owners face similar challenges. When they are unavailable, potential customers may wait too long for a response, receive inconsistent information, or leave without taking the next step.

This helped us understand that a useful AI product must do more than generate impressive answers. It must understand the business, communicate its message consistently, and help convert customer interest into action.

Using Vertex AI, Gemini 2.5 Flash, and the text-embedding-004 model taught us how Retrieval-Augmented Generation can help small businesses answer customer questions using their approved documents. This is especially important for brand coaches because their advice, programmes, prices, processes, and brand voice may be unique. The AI must represent that information accurately instead of providing generic or invented answers.

We also learned that one of the biggest challenges for small businesses is managing leads while handling everyday operations. Connecting Google Calendar and Google Meet showed us how an AI agent can answer a visitor’s questions, understand their needs, check the business owner’s availability, and book a consultation without manual back-and-forth. This reduces the possibility of interested customers being forgotten or lost.

Building the voice experience with Google Cloud Speech-to-Text and Text-to-Speech showed us that accessibility matters. Some visitors prefer speaking to typing, especially when using a mobile device or explaining a complex need. Voice allows small businesses and brand coaches to serve more people in a natural and convenient way.

We also learned that content creation is a continuous challenge. Small businesses often know what they want to communicate but may not have the time or budget to hire a full marketing team. Gemini can help turn their business knowledge and customer conversations into relevant content ideas and drafts while keeping the business owner in control of the final message.

Most importantly, we learned that AI should support people rather than replace them. Small-business owners and brand coaches still provide the expertise, creativity, empathy, judgment, and relationships that make their businesses valuable. EnsightLabs helps them handle repetitive work, respond faster, remain consistent, and focus more of their time on serving customers and growing their businesses.

What's next for EnsightLabs

Our immediate priority is to learn from our paying customer and the other businesses testing EnsightLabs. We want to collect honest feedback, understand how the product performs in real business situations, and use those insights to improve its reliability, speed, usability, and overall value.

We plan to make onboarding simpler so that a small-business owner or brand coach can create an agent, upload business information, customise the experience, and add it to a website without needing technical knowledge. We will also improve the tools for reviewing conversations, updating the knowledge base, measuring lead quality, and understanding which interactions result in meetings or sales opportunities.

On the AI side, we plan to improve the grounding and reasoning capabilities powered by Vertex AI and Gemini. We also want to expand multilingual voice support using Google Cloud Speech-to-Text and Text-to-Speech so businesses can serve customers who prefer different languages or find speaking more convenient than typing.

We will continue improving the Google Calendar and Google Meet workflow while introducing deeper integrations with the tools businesses already use for customer management, communication, and marketing. Our goal is for EnsightLabs to support the complete journey from a customer’s first question to qualification, booking, follow-up, and conversion.

Content creation and financing readiness will remain important parts of the platform. We want to help small businesses and brand coaches turn their knowledge and customer conversations into consistent marketing content while also helping them understand how prepared they are for financing opportunities.

From a business perspective, our next goal is to reach more paying customers through product demonstrations, partnerships, organic content, referrals, and customer success stories. As EnsightLabs grows, we also hope to create opportunities for people working in customer onboarding, content strategy, AI operations, integrations, and business support.

Our long-term vision is to make practical and affordable AI accessible to small businesses—not as a replacement for their expertise, but as a reliable growth partner that helps them respond faster, operate more efficiently, and serve more customers.

Built With

Share this project:

Updates