Applied Labs is now generally available

Every support leader can name the workflow that broke the old model.

The delayed order that required carrier data. The cancellation that needed subscription history, offer limits, and a little empathy. The VIP escalation buried in the same queue as password resets. The product bug customers described in 50 different ways before anyone on the product team saw the pattern.

The first wave of AI support tools mostly looked like better chatbots: answer a question, link an article, maybe deflect a ticket.

That is useful. It is not enough.

The hard work starts when a customer needs something resolved. The agent has to understand the customer, read the right systems, follow policy, take the approved action, and hand off cleanly when judgment belongs with a person. It has to do that across chat, email, voice, SMS, social, helpdesk queues, retention flows, conversion moments, and the messy long tail of internal systems.

We started Applied Labs because support teams should not have to choose between automation that scales and service that customers trust. Today, Applied Labs is generally available.

Teams can now start from public plans, run a 14-day free trial, and deploy the same platform our customers use to operate AI agents, human support teams, helpdesk, CRM, analytics, and governed CX operations in one place.

Start a free trial, or talk to Applied Labs about the AI agents your team wants to launch next.

What you can do now

The launch is simple: start a trial, connect the first sources your team trusts, and test an AI agent on real customer work.

In the first 14 days, teams usually pick one high-signal starting point:

1. A high-volume support queue where customers ask repeatable questions but still need order, subscription, account, or policy context.
2. An agent that needs to do more than answer because a person still has to update an address, check eligibility, issue a make-good, create a handoff, or route an exception.
3. A retention or conversion moment where the conversation needs customer history, offer rules, product context, and a measurable outcome.
4. A helpdesk migration path where AI-handled conversations, human replies, customer memory, CRM fields, QA, and analytics should live together.
5. A trust and quality problem where the team needs traces, audits, QA scorecards, and analytics before expanding automation.

The goal is not to automate everything in week one. The goal is to make one real AI agent observable, controlled, and measurable.

Why GA now

GA is not a tagline for us. It means the platform is ready for a much wider set of teams.

That readiness shows up in five ways:

• Public plans and self-serve trials: teams can now price, trial, launch, and expand without starting from a bespoke enterprise conversation.
• Production agent coverage: customers are already using Applied across chat, email, voice, SMS, community, support queues, cancellation flows, conversion agents, proactive agents, and customer intelligence.
• Operational control: knowledge references, reasoning visibility, approvals, read/write permission tiers, audit trails, QA, and safe failures are part of the product, not afterthoughts.
• Enterprise-ready governance: Applied supports SOC 2 Type II, HIPAA, GDPR-aligned controls, least-privilege access, RBAC, static IP, secrets, procurement support, and SLAs for teams that need them.
• A clear expansion path: start with one queue or agent, validate the traces, then add channels, agents, connectors, analytics, and custom agent actions as the program matures.

Over the last year, we worked closely with customer operations teams to learn what the product needed to be. General availability is the milestone where that learning becomes a public platform: teams can now trial Applied Labs, launch a first AI agent, and expand from published plans.

Pick your starting point

You can now choose the plan that matches your operation:

• Launch for teams moving their first support queues into Applied helpdesk and CRM.
• Growth for support teams automating repeat issues across growing queues, including Save Agent, Conversion Agent, voice, SMS, database connectors, and HTTP actions.
• Scale for high-volume teams running multiple queues, brands, and AI agents, with Proactive Agent, traffic control, Quality Assurance, Insights, custom dashboards, and custom analytics.
• Enterprise for flexible volume pricing, custom connectors, custom help desk integrations, help center workflows, procurement support, static IP, RBAC, secrets, and SLAs.

Every plan is designed around a practical promise: start small, prove outcomes, and expand only when the team can see what changed.

Why we built this

Support software was built around inboxes. That made sense when the job was to route, assign, and close tickets.

But the customer conversation has become the front line for almost everything else:

• fulfillment issues,
• billing and subscription problems,
• cancellation saves,
• conversion objections,
• product bugs,
• policy confusion,
• VIP escalations,
• and early signals that something in the business is breaking.

If those conversations stay trapped in an inbox, the company learns too slowly. If an AI agent only answers FAQs, the customer still waits. If an agent can act without control, the business will not trust it.

So our product principle has been simple: an AI agent in customer operations has to do three things well.

1. Answer with source-grounded knowledge and brand voice.
2. Act in the systems where work actually gets done.
3. Escalate with enough context that the human does not restart the investigation.

Everything in Applied Labs is built around that loop.

What is Applied Labs?

Applied Labs is one platform for AI Agents, helpdesk, and CRM.

It gives teams the pieces they need to run customer-facing AI in production:

• AI Agents for support, retention, conversion, proactive outreach, and customer operations.
• Help desk + CRM so AI-handled conversations, human replies, escalations, customer history, and ownership live in one operating queue.
• Omnichannel coverage across chat, email, voice, SMS, social, and more.
• Connectors and custom agent actions so agents can read from and act in tools like Shopify, Zendesk, Gorgias, Stripe, Recharge, Airtable, internal APIs, and databases.
• Control Center for knowledge, approved responses, routing, escalation policy, and agent logic.
• Evaluation, analytics, and Insights so teams can measure AI resolution, CSAT, quality, intent mix, handoffs, workforce impact, and the product issues hiding in conversations.
• Security and governance with least-privilege access, human review, audit trails, SOC 2 Type II, HIPAA, and GDPR-aligned controls.

The point is not to make AI sound better in a chat window. The point is to resolve real customer work while keeping teams in control.

How teams are using Applied Labs today

The best way to understand Applied Labs is to look at the AI agents and CX programs customers are already running in production.

"With Applied Labs, our bot now outperforms human agents on CSAT, and automation has nearly doubled - all while cutting our platform costs in half."

Caitlin Logan

VP of Customer Service, FabFitFun

This is the pattern we see across customers: the AI does not remove the human team. It removes the work that kept humans from doing the work customers actually remember.

The AI agent shift: from "reply faster" to "resolve the issue"

A typical AI support flow looks simple from the outside.

A customer asks, "Can you update my shipping address?"

Before Applied, that request often looked like this:

1. A bot explains the policy.
2. The customer waits.
3. A human opens the ticket, checks the order, verifies shipment status, edits the address, and replies.
4. If the order already shipped or the address is invalid, the human starts a second branch of the workflow.

The bot made the first response faster, but the work still landed with a person.

With Applied:

1. The agent understands the request and retrieves the current order.
2. It checks the address-change policy and eligibility.
3. It uses a permissioned connector to update the allowed fields, or routes the exception to a human with the full reason.
4. It confirms what changed, records the trace, and writes the outcome back into analytics.

That is the difference between an answer and an outcome.

The same pattern applies to refunds, replacements, subscription pauses, cancellation saves, delivery recovery, failed payments, VIP routing, product-fit questions, renewal risk, and proactive outreach.

How it works

Applied AI agents follow the same operating loop whether they start from an inbound message, a phone call, a cancellation moment, a conversion question, or a proactive signal.

1. Understand the request or signal
   Identify the customer intent, risk, channel, and agent type.

2. Load the right context
   Retrieve customer memory, CRM data, order state, subscription facts, policies, past conversations, and relevant knowledge.

3. Apply policy and controls
   Check eligibility, source coverage, offer limits, permissions, consent, frequency rules, and escalation criteria.

4. Take the approved path
   Answer, draft, update a system, create a handoff packet, route a review, send outreach, or stop when the safe path is unclear.

5. Record the trace
   Keep the sources, decisions, actions, handoffs, failures, and outcomes available for review.

6. Measure and improve
   Feed the outcome into analytics, QA, Insights, and the next agent release.

This is why we keep talking about operations instead of chatbots. The product is the loop.

Trust and control are the product

Customer-facing AI cannot be a black box. It has to be operated.

Applied Labs gives teams control at each layer:

• Knowledge references show which sources shaped the answer.
• Knowledge gaps identify where the agent lacked coverage.
• Reasoning visibility helps teams understand why a response happened.
• Least-privilege connectors limit what the agent can read or change.
• Read, write, and sensitive-write tiers separate low-risk context from high-risk actions.
• Human review and escalation paths keep judgment with the team when policy is ambiguous.
• Audits and QA scorecards evaluate every conversation, not just survey responses.
• Analytics and Insights turn conversation data into product, CX, and operations decisions.
• Safe failures stop, retry, answer from knowledge, or escalate rather than guessing.

We do not believe teams should "set and forget" customer-facing AI. The best agent programs run like product programs: test, measure, improve, and ship the next controlled release.

Getting started

If you are starting from an existing helpdesk, the first AI agent usually begins with one high-volume queue. Connect the knowledge sources and systems of record, run the agent in review or limited traffic, audit the traces, then expand once the team trusts the behavior.

If you are replacing a helpdesk, Applied can become the operating queue for both AI and human work. Conversations, customer memory, CRM context, handoffs, and QA live together instead of splitting across separate tools.

If you are already running AI, the highest-leverage starting point is often the gap the current system cannot close:

• AI answers but cannot act.
• AI resolves easy tickets but fails when an agent needs deeper context or action.
• AI cannot connect to internal data.
• AI works in chat but not voice, email, SMS, or social.
• AI metrics exist, but no one trusts the readout.
• Support has the customer signal, but Product and Operations cannot use it.

GA makes that first step easier. Start with a trial, use the free AI resolutions to test a real agent, and move toward a plan only when the outcome is visible.

FAQ

Do I need to replace my helpdesk?
No. You can start from an existing helpdesk, connect the systems and knowledge sources the agent needs, and test one queue or agent first. If you want Applied to become the operating queue, Applied helpdesk and CRM can bring AI-handled conversations, human replies, customer memory, handoffs, QA, and analytics together.

What can I realistically test in 14 days?
Pick one concrete AI agent: order status and delivery recovery, address changes, refund or replacement eligibility, cancellation saves, product-fit questions, VIP routing, or a queue where the current AI answers but cannot act. The best trial proves whether Applied can resolve or route that work with the right context, controls, and trace quality.

Do I need engineering support to get started?
Not for every starting point. Many teams begin with knowledge sources, policies, helpdesk context, and standard connectors. Engineering helps when the first AI agent depends on internal APIs, databases, custom systems, or sensitive write actions.

Which actions are safe by default?
The safest rollout starts with reads, drafts, review mode, and escalation. Write actions can be added through permissioned connectors, with separate read, write, and sensitive-write tiers, human review paths, and audit trails.

What data does the agent use?
Only the sources and systems your team connects and permits: knowledge articles, policies, past conversations, CRM records, order or subscription state, helpdesk context, approved responses, and custom data sources. Trust controls show the references, decisions, actions, handoffs, failures, and outcomes behind the agent's behavior.

When should I talk to sales instead of starting self-serve?
Talk to us if you need custom connectors, help desk migration planning, multiple brands or queues, procurement support, static IP, RBAC, secrets management, SLAs, HIPAA-sensitive deployments, or flexible volume pricing. Start self-serve if you want to test a focused AI agent before expanding.

What's next

We think customer operations is moving from reactive queues to governed agent systems.

Support agents will keep resolving inbound issues. Save agents will turn cancellation moments into measurable retention programs. Conversion agents will help shoppers while intent is still warm. Proactive agents will act on signals before customers have to ask. Analytics, QA, and Insights will turn every conversation into the next improvement.

The long-term goal is simple: every customer, better served.

Applied Labs is generally available today.

Start a free trial, or talk to Applied Labs about the AI agents your team wants to launch next.