Case Study: Building the AI Operating System for Modern Commercial Teams
How Lithe co created a new era of Go To Market engineering:
A new challenge came to us, was simple to state and hard to fix. Commercial operations had outgrown the systems. Data lived in silos. Marketing used one set of tools, sales used another, everything was tracked separately, and leadership lacked a single source of truth for consultancy sales operations.
Teams were smart and motivated across multiple regions. Work still happened by copy and paste. Leads moved from spreadsheets to CRMs, from emails to chat apps, from follow ups to follow laters. Both people and systems where lagging.
So, we set out to build an intelligent layer that connects people, data, and AI into one living process. We call this Go-To-Market engineering / GTM-engineering.
Where it started
We began with tools already in use. Airtable became the heart of the system, a living database for leads, conversations, and relationships.
Automation came next. Using Make dot com and n8n, we wired connections between apps such as LinkedIn, Gmail, HubSpot, and internal CRMs. Incoming messages triggered updates automatically. Status changes in Airtable created reminders and tasks.
Then we brought in various large language models. Instead of treating AI like a chatbot, we embedded it into the workflow. When a message arrived, an AI prompt interpreted it, classified intent, and drafted a personal response for human approval.
Within weeks, follow up speed had tripled. Copy and paste work disappeared.
Turning automation into a platform
The prototype proved the point. To scale across thousands of leads and multiple teams, we rebuilt the core in NodeJS with a modular architecture. Each automation became a service with its own logic and API. Prompt logic was turned into reusable functions that can run at scale.
The front end experience became a single mission pipeline that showed every contact, conversation, and relationship in real time. The system did not only automate tasks. It also revealed how strong or weak relationships were becoming over time.
Stack overview
NodeJS and TypeScript for application logic
PostgreSQL for structured data
Vector storage for semantic search and relationship scoring
API integrations for LinkedIn, WhatsApp, Workspace / Email, and a major CRM
LLM orchestration using OpenAI, Agentic Claude, and optional local models for privacy sensitive clients
Governance and compliance were built in from day one. All data ran under the client’s credentials within their cloud. GDPR alignment was part of design, not an afterthought.
Engineering principles behind it
The goal was not to automate everything. The goal was to automate flow.
Embed AI, do not bolt it on. Prompts run inside the process automatically. No one needs to paste text into a chatbot.
Keep humans in the loop. Every message and action can be reviewed, approved, or edited before sending.
Measure relationships, not clicks. The true indicator is engagement and response quality, not only volume.
By connecting data, communications, and AI in one platform, the organisation could finally see what was happening across commercial work in real time.
From manual to intelligent flow
Before
Each team worked in its own system
Follow ups were inconsistent and often missed
Reporting required manual updates
Response time lagged behind competitors
After
AI workflows handled
Relationship health became visible across all channels
Reports generated themselves every 24 hours
Pipeline velocity increased across regions
In practical terms, time spent on repetitive outreach and updates fell by about sixty five percent. The same people delivered three times the output with higher quality and less fatigue.
Beyond automation: real Go To Market engineering
GTM engineering blends software development, AI delivery, and process design. A GTM engineer understands business flow and system architecture. They know where data starts, where it goes, and how to turn it into action. They design prompts, connect APIs, keep data clean, and build dashboards that show performance as it happens.
We co developed in the open with sales and marketing leaders. Each iteration was shaped by real usage and side by side delivery. Engineers and commercial strategists worked together until technology felt human again.
Inside the build
Data layer Airtable was the first foundation. We later moved to PostgreSQL with custom schemas for control and performance.
Automation layer Make dot com handled early integrations. As logic grew, we built native connectors in Node dot js and n8n for speed and flexibility.
AI layer LLMs powered interpretation and writing. Prompts were modular, parameterised, and version controlled. The model learned the company tone over time.
Interface layer The product visualised every ongoing conversation across channels. Sales and marketing leaders could approve messages or adjust tone in seconds.
Analytics layer Vector search and metadata tracking produced live dashboards on response rates, sentiment, and relationship growth.
The impact
Teams stopped switching between tools and worked inside one unified mission pipeline. AI handled repetition. People handled relationships.
For leadership, visibility changed the game. They saw where opportunities were heating up, which relationships needed attention, and where engagement was strongest.
For delivery teams, the system acted like an always on partner. It suggested follow ups, wrote drafts, and captured learning continuously.
Results
Three times faster outreach cycle
Seventy percent fewer manual tasks
Full data visibility from first contact to deal
A central AI operating layer ready to scale globally
Lessons learned
Start small and connect what already exists. Airtable and Make dot com are powerful starting points.
Let AI assist, not decide. Keep humans in the approval loop.
Move to code when ready. No code prototypes speed discovery, custom architecture brings control and scale.
Design around people. Good systems fit how teams actually work.
What this means for modern companies
Most organisations have the same root problem. Too many tools, too many handoffs, not enough flow. GTM engineering treats go to market as a product that can be designed, automated, and scaled like software. AI delivery makes it faster, clearer, and more intelligent.
The outcome is not only efficiency. It is a new rhythm of work where technology quietly clears the path so people can focus on value.
Ready to build your AI operating layer
Lithe Transformation helps organisations rethink how commercial systems work. Whether you are starting with a no code setup or scaling to full stack automation, we design, build, and embed the technology that makes growth flow.
If your teams are still copying, pasting, or waiting for data to move, book a meeting with Abraham Schoots to talk to us about Go To Market engineering today.