The AI support startup whose product pitch is, essentially, that everyone else is fibbing about their resolution rates - and it has the metrics to prove otherwise.
There is a genre of software company that sells you a number, and the number is customer-support resolution rate, and the number is usually somewhere north of 80 percent, and if you ask how the number was calculated you tend to get a change of subject. This is not necessarily fraud. It is more like a shared industry understanding that the resolution rate is a marketing artifact - a thing that goes on a slide - rather than a claim you are supposed to audit. Everyone knows this. Almost nobody says it out loud.
Applied Labs, a New York company founded in 2024, says it out loud. Its co-founder and chief executive, Michael Woo, has been on the record being what he calls skeptical about the resolution rates that better-known providers advertise, which is a bracing thing to build a company on, because it means your own numbers had better be honest or the whole pitch collapses. So the company built the honesty in. Its analytics, per its own description, are designed to offer "more honesty and visibility about what AI is really resolving" - the theory being that if you are going to compete on trust in a category full of inflated claims, the measurement is the product.
What Applied Labs actually sells is more prosaic and more useful: on-brand AI agents that handle customer support and operational workflows across every channel a customer might turn up in - chat, email, phone, SMS, Slack, Instagram, Facebook. Each agent is trained on a company's own knowledge base and sounds like that company rather than like a generic chatbot. The agents come in flavors - Support, Save, Conversion, Proactive, Assistant - and sit on top of an AI-native help desk, a CRM, a voice system and an analytics layer that the company rebuilt from scratch rather than bolting onto legacy tooling.
The framing the company keeps returning to is that AI should not replace the support team so much as amplify its best instincts. Woo's line is that "you get the best results when AI efficiency is combined with human judgement" - efficiency where you want scale, humans where you want trust. It is a less thrilling story than "fire your support department," and it happens to describe how the good deployments actually work: the machine takes the repetitive volume, the person keeps the calls that need a person.
"The bottleneck is no longer the model. Quality, speed and cost have reached an inflection point."
- Michael Woo, Co-Founder & CEO, Applied LabsThe reason to take the skepticism seriously is the resume behind it. Applied Labs' founders are not first-timers narrating AI from a distance - they built it. Woo was employee number 20 or so at Scale AI, where he led a team of 30 focused on operational scalability. His co-founder, Soham Waychal, previously led engineering at the a16z-backed commerce startup Canal and holds five AI patents. The idea for the company came directly out of their Scale AI years, watching how cumbersome and repetitive support and operations work really is - and noticing that this was exactly the shape of problem AI had gotten good at.
Early Scale AI employee (~#20) who led a 30-person operational scalability team. The company's public voice on why AI support metrics should be audited, not admired.
Former engineering lead at a16z-backed Canal. Holds five AI patents. The technical half of a founding team that had built agent systems before naming a company after them.
The pitch to a support or operations leader is straightforward: deploy an agent trained on your knowledge base, let it self-learn from each interaction, and watch it resolve the repetitive volume - order status, returns, "where's my refund" - while routing the genuine judgment calls to humans. It plugs into the tools you already run: Shopify, Stripe, Zendesk, Salesforce, Slack, Airtable, Notion and more than a dozen others.
Support, Save, Conversion, Proactive and Assistant agents - on-brand and omnichannel, across chat, email, voice, SMS, Slack and social.
An AI-native help desk with a unified inbox, smart routing and unlimited seats.
AI-native customer relationship management, built to work with the agents rather than beside them.
Automated voice support that carries a phone conversation from hello to resolution.
Metrics designed for honesty - visibility into what the AI is genuinely resolving, not what looks good on a slide.
Set up, authorize, test and monitor agents in one place - run and evaluate them before they touch a customer.
Here is the part that makes the skepticism structural rather than rhetorical. Most AI support tools charge per seat, which means you pay whether or not the agent resolves anything. Applied Labs charges on outcomes - you pay for results, not for seats that sit idle. It is a pricing model that only makes sense if you are confident your agents actually work, which is a nice way of forcing your incentives to line up with your customer's. The company wraps that with a 90-day risk-free opt-out and what it describes as zero-downtime migration, the two things that most often stop a support team from switching vendors at all.
The early evidence is a customer list heavier than a one-year-old company usually carries: Sundays for Dogs, FabFitFun, Truemed, Warren Lotas, Nabis, ripple+, Smalls, GHOST and Lemonade among them. FabFitFun's Caitlin Logan is quoted saying the company's AI-agent CSAT hit 95 percent while automation "nearly doubled." Roughly a dozen customers signed in year one, with revenue growing fast enough to bring in real institutional money.
The security posture is the enterprise table-stakes version of the same trust story: SOC 2 Type II, HIPAA and GDPR-aligned. None of which is glamorous, all of which is the price of admission for handling other companies' customer conversations at scale.
"AI allows you to scale up your best human judgement on an infinite volume of tasks."
- Michael Woo, Applied LabsIn January 2025, Applied Labs announced a $4.2 million seed round led by Abstract, with Point72 Ventures, Outlander and Tetra joining, plus angels including the former Twitter executive Ali Rowghani. That brought total funding to $5.2 million. The stated use of proceeds was the least surprising thing in the announcement: hire more engineers.
It is worth noticing what the investors were buying. In a category where the differentiator is usually a bigger claimed resolution number, Applied Labs raised money on the opposite - a promise to report smaller, truer numbers, and to charge only when the work gets done. That is either a durable wedge or a hard way to sell software. The customer list suggests the market has at least some appetite for the former.
The competitive set is crowded and well-funded - Intercom's Fin, Sierra, Decagon, Ada, Gorgias, Zendesk AI, Salesforce Agentforce. Applied Labs' answer to all of them is the same: not a bigger number, a truer one, priced so you only pay when it lands.
Applied Labs is a New York-based AI company building on-brand, omnichannel AI agents that handle customer support and operational workflows across chat, email, voice, SMS, Slack and social channels. Founded in 2024 by former Scale AI operators Michael Woo and Soham Waychal, the company pairs AI efficiency with human judgment and pitches honest, verifiable resolution metrics over the inflated claims common in the AI support market. Its platform spans AI agents, an AI-native help desk, CRM, voice and analytics, sold on outcome-based pricing rather than per-seat fees.
Last updated: