Building an AI Agents Platform With $2M Funding

AI Agents Platform With $2M Funding

Powering a new-generation AI workspace that liberates consultants from routine work

Project Highlights

Full product redesign shipped across every major section of the platform

Modules running in production for 6+ months with zero major incidents

New integrations added in under 2 weeks thanks to modular architecture

DocAssist shipped for both Google Workspace and Microsoft 365 from a shared core

$2M seed round closed during the engagement

Client: Moonnox (Chicago, USA)

Industry: AI / B2B SaaS

Engagement: Ongoing

Uinno Team: Engineering Manager, Project Manager, Frontend Developer, Backend Developer, AI Engineer, Product Designer, QA team

See it live: Moonnox

3 levels of AI agents

L1: System Agents

Pre-configured by Moonnox and locked at the platform level. Even admins cannot change them. These handle the most critical workflows, like the "golden" call summary agent that produces consistent meeting summaries across every organization.

‍

L2: Template Agents

Users choose from a library of ready-made agents such as Kickoff, Retrospective, Implementation, and Drafter. Each comes with a working prompt that users can customize to fit their team's language and process.

‍

L3: Custom Agents

Built from scratch. The user starts with a blank prompt, picks the model, sets advanced parameters, and connects the agent to specific project contexts. Full control for teams that know what they want the AI to do.

‍

TECH DEEP DIVE

Functional Design

The product had grown fast, but the initial UI was not user-friendly.

Solution

We redesigned every major surface of the platform piece by piece, shipping new interfaces alongside feature work so customers never experienced downtime or a jarring overnight change.

Structuring Startup Chaos

Like many startups shaping a new category, requirements shifted constantly as the founders learned what drives value. Ideas came fast, but they needed structure before they could become working features.

Solution

Uinno turned founders ideas into actionable development requirements with clear scope, priorities, and timelines, keeping the team productive through constant change instead of waiting for specs that would never be final.

Collaboration with the In-House Team

As the product grew, two teams working on the same codebase needed shared standards to avoid breaking each other's work.

Solution

We aligned development workflows with the Moonnox team by establishing shared quality benchmarks and a synchronized PR review process, keeping the system robust and scalable as the codebase expanded.

What Uinno Built

Module 1. The Event-Driven Suggestion Feature

When a client meeting wraps, Moonnox processes the call against the project context stored in its database. We built the "Suggestion Service" as an abstract, event-driven "black box".

The Suggestion Feature needed to process events from multiple systems, understand what happened in a meeting, and reliably generate the right action suggestions for the right context, every time, at scale. It evaluates the context and proposes next steps like creating a Jira ticket, drafting a email follow-up with attachments, or posting to Slack.

Because we built integrations as self-contained modules, the system is infinitely scalable. That is why adding a new action type takes under two weeks without touching core code.

Module 2. Co-Pilot and Live Coaching

Co-Pilot module joins client calls and works as an active participant, not just a recorder. Beyond providing post-call transcripts and sentiment analysis, it offers Live Coaching during the meeting itself.

If a consultant is pulled into a large call mid-way or loses track during a complex discussion, the Co-Pilot provides real-time mini-summaries of what has been said in the last few minutes, directly on screen.

The AI processes the conversation as it happens, pulling in relevant project context from the knowledge base so the summaries are not just a recap of words spoken but a contextual briefing tied to the client, the engagement, and the decisions that matter.

Module 3. DocAssist for Google and Microsoft

DocAssist brings Moonnox agents and project context directly into Google Docs, Sheets, Slides, Word, Excel, and PowerPoint. The AI draws from everything stored in the Moonnox knowledge base, including past calls, project documentation, and connected sources like Confluence and Google Drive, so the content it generates is grounded in real project data, not generic output.

A consultant can highlight a section of a proposal, point at an agent configured for that client, and get a draft that already reflects the right terminology, project history, and deliverables without switching between tools or copy-pasting from old files.

Module 4. Project Management

Project management layer allows users run their engagements without paying for a separate tools like Jira. But unlike a standard task board, this module is connected to the same AI context layer that powers everything else in the platform.

A Project Management head can open a dashboard and see health scores across all active engagements, generated through sentiment analysis of what is happening in calls, tickets, and team communication.

Module 5. Project Health Scoring

What’s Next: Expanding the Agentic Ecosystem

Advanced Agent Skills & Chains

We are developing an advanced builder that allows agents to execute complex chains of actions. For example, an agent will be able to search the internet for specific information, summarize it, cross-reference it with internal project context, and output a highly tailored result.

Agent Onboarding Wizard

To help users get the most out of custom L2 agents, we are creating a step-by-step onboarding flow. This will guide new users to upload the right context such as connecting Jira or uploading key documents from the very start, to ensure the agent performs optimally.

AI-powered Test Automation

Agents generating and maintaining automated test coverage, cutting manual QA load and making releases faster.

Preparing the platform for higher load

As Moonnox grows its client base, Uinno is helping plan the scaling work that comes with serving more organizations and larger teams inside the product.