An applied AI studio, running on its own work.
The setup
Their team kept shipping — cloud AI, on-prem LLMs, custom integrations. The agents ran everything around it.
UCLab is an Ottawa applied AI studio building software, automation, and private LLM systems for businesses that need to move fast and keep their data close. Their team wanted to spend their days designing systems — not triaging inbound, qualifying leads, or rewriting the same proposal for the fifth time.
The work itself was never the problem. The work around the work was. Ten to thirty inbound conversations a week doesn't sound like a lot — until you trace what each one actually demanded. First-touch qualifying. Research on the prospect and their industry. Writing the proposal. Drafting the SOW. Following up. Coordinating internally on who's doing what. Six categories of work, all pulling on the same two founders, every week.
A single proposal, end to end, took a week. Not a week of focused effort — a week of calendar time, scattered across other commitments, with the prospect waiting between each step.
Karmaflow gave them a working layer underneath the studio — one that could carry the operational weight without the team having to scale headcount to match.
What got built
Three layers. Day one to done in three weeks.
The deployment wasn't a single agent or a single integration. It was a three-layer operating system stacked underneath the studio — built in phases. Day one went live with web chat and a deliberately limited goal: qualify inbound, hand off cleanly. Three weeks later, the full system was in production, including the layer that surprised them most.
Omnichannel customer agent
Live across web, voice, SMS, and email. Qualifies and triages inbound, answers pricing and scoping questions, books calls into the founders' calendars, captures discovery context, and hands off to a human when needed. Same voice, same context, every channel.
Per-person operational agents
Six to ten people inside the studio, each with their own agent. Research and prospect briefing, proposal and SOW drafting, contract generation via DocuSign, email drafting and follow-ups, CRM updates, internal Q&A against the knowledge base. Wired into Zoho, Google Workspace, QuickBooks, web analytics, and the studio's internal KB.
Scheduled agent, daily sweep
A dedicated agent on a schedule, running a daily risk-and-opportunity assessment across every data source the studio touches — email, meetings, CRM, web chat, every agent and human exchange. Reviews the CRM daily and acts: follows up internally and externally as needed. Nothing slips because nothing waits to be noticed.
What changed
A week of work. Ten minutes.
The biggest shift wasn't a percentage — it was a category change. The chain that used to take a week now runs end-to-end in about ten minutes. A week is something you go home with. Ten minutes is a coffee.
First-touch resolution over the last 90 days. Conversion roughly doubled — ~10% to ~20% — with the same team and no new hires.
One Monday morning · anonymized
The studio was asleep. The agent wasn't.
A deep-tech prospect in nano-optics, based in Germany, opens the web chat at 4:00 AM Ottawa time. What follows is indistinguishable from a fully-staffed firm answering in real time — except no one is awake.
- 4:00 AM
Prospect lands on the site. Conversation begins.
- 4:20 AM
20-minute chat. Agent qualifies as Senior SDR-fit, moves the lead MQL → SQL, hands off to the operational layer. Research kicks off; a personalized package is built — case studies, answers, NDA.
- 4:30 AM
Package and NDA delivered. Ten minutes from end of chat.
- 5:00 AM
Prospect signs the NDA via DocuSign. SOW agent drafts the statement of work; sweeper posts a handoff bundle to Slack — transcript, research, NDA, context — tagged to the right humans.
- Same day
Team walks in to a primed prospect. SOW signed.
What's next
The studio kept growing. The org chart didn't.
UCLab built this for themselves first, which is the part that matters. An applied AI studio that runs on a working version of what it sells. The team isn't replacing the work — they're running a version of the firm they couldn't have staffed.
What's next on the platform: expanding the operational agent layer to more roles inside the studio, and bringing the same approach to marketing automation — the next category of work pulling on the team's time.
“We planned for 20–30% autonomous resolution. The outcomes were exponential.”
— Founders, UCLab.ai
About UCLab
An applied AI studio in Ottawa building software, automation, and private LLM systems for businesses that need to move fast and keep their data close. Founded 2020.
VisitAbout Karmaflow
The AI agent platform underneath. Businesses use Karmaflow to build and launch agents across voice, SMS, email, and chat — and orchestrate them as a working operational layer, not a collection of chatbots.
Start a conversation.
See what a three-layer agent operating system looks like running underneath your team.
- Case Study
- Customer Story
- UCLab
- AI Studio
- Applied AI
- Autonomous Workforce
- Per-person Agents
- Ottawa