What This Actually Is and Why It Exists
Every team has a short list of tools that carry the real weight of the operation. An applicant tracking system for hiring. A CRM for managing the sales pipeline. An issue tracker for engineering. An e-commerce platform running the actual business. These tools are not the problem. They work. The problem is everything that needs to happen around them to keep them current, accurate, and actually useful.
Checking whether a candidate has moved stages. Following up on a deal that has gone quiet. Surfacing an unassigned engineering ticket before it gets buried under three days of newer work. Flagging a delayed shipment before a customer complaint arrives. None of this is difficult work. But it is relentless, repetitive, and easy to quietly deprioritize when other things feel more urgent, right up until the moment it becomes a real problem that someone has to spend real time untangling.
The Agents marketplace exists to take that category of work off your plate entirely. Think of it as a shelf of pre-built automations, each one wired directly into a service your team probably already uses, ready to deploy without requiring an engineer to build a custom integration or a consultant to configure a brittle third-party workflow tool that will need maintenance every time one of the connected services updates its API.
You find the agent built for the tool you actually use, connect it to your account, and it starts working inside that tool immediately. No custom development, no complex configuration, no lengthy onboarding process. What makes this meaningfully different from most automation platforms is that you never hand off blind control in exchange for the convenience of automation. Every agent can be paused the instant you want it to stop, deleted entirely if you decide it is not useful, and reviewed in precise detail to see exactly what it has done, without digging through nested settings menus or waiting on a support ticket.
Why Seven Agents and Not Seven Hundred
A marketplace with hundreds of integrations sounds comprehensive until you actually try to use one and discover that most of them barely work, each one doing one narrow thing in a way that was never tested against how teams actually operate. A long list of shallow integrations is not a feature. It is a maintenance liability dressed up as a catalog, and it erodes trust quickly the first time an agent does something wrong because it was never built with real depth behind it.
Every agent in this marketplace was built to genuinely understand the tool it connects to, not just fire off a generic API call and log a success. That takes real time, real testing against real workflows, and real iteration based on how teams actually use the tool in practice rather than how the API documentation suggests it should be used.
Seven agents exist today because seven agents have been built to that standard. We would rather add an eighth agent that meets that bar than rush to twenty that do not. As genuine demand emerges from the teams using Kira for a new integration, that is exactly how the next one gets prioritized, based on what real teams actually need, not by chasing a number to display on a marketing page.
What the Marketplace Gives You
A Browsable Catalog of Agent Cards
Every agent is presented as a clear, recognizable card tied to a specific brand and service. You never have to guess what an agent actually connects to before you decide to deploy it. The catalog is designed to be scanned quickly so you can find what you need and get it running without unnecessary friction.
One Click to Run, One Click to Pause, One Click to Remove
Nothing about this system locks you into a commitment you cannot easily walk back. Deploying an agent takes a single action. Pausing it takes another. Removing it entirely takes one more. At no point are you navigating a complex interface or submitting a request to disable something that is running on your behalf. Control is always immediate and always yours.
Live Status on Every Agent
You know definitively what is active right now, not what was active five minutes ago based on a dashboard that refreshes on a schedule. The status indicator updates in real time so there is never a guessing game about what is currently doing something on your behalf versus what is sitting idle. A lot of automation tools treat status as an afterthought, something you check if you remember to look for it. Here it is front and center, because trusting an agent with real actions requires knowing exactly what it is doing at any given moment.
A Complete Activity Log
Every action an agent takes lands in the activity log the moment it happens, without exception. This is not a summary generated after the fact that smooths over details or rounds off the edges of what actually occurred. It is a granular, sequential record of exactly what happened and exactly when, creating a genuine audit trail rather than a vague account you have to piece together after the fact.
Real Time Toast Notifications
Because a log sitting in a dashboard is easy to forget about entirely, we also surface a notification the moment something meaningful changes. You catch important activity as it happens rather than discovering it three days later while scrolling through history trying to reconstruct a sequence of events. The combination of live status, a real time activity log, and immediate notifications means automation never has to mean losing track of what is actually happening on your behalf.
What Is in the Marketplace Right Now
Each of the seven agents currently available was built with real depth for the specific tool it connects to. Here is what each one does in practice, not in the abstract.
| Agent | Connects To | What It Looks Like in Practice |
|---|---|---|
| Greenhouse | Recruiting and ATS | A candidate moves to the interview stage and the agent automatically nudges the hiring manager, flags a scheduling gap, and updates the pipeline dashboard so nobody has to remember to check manually |
| Shopify | E-commerce | An order flags as delayed shipping and the agent surfaces it before a customer complaint arrives rather than after, giving your team time to get ahead of the situation |
| PagerDuty | Incident Response | An alert fires at 2am and the agent tracks acknowledgment time and escalates automatically if nobody responds within the window you have configured |
| HubSpot | CRM and Marketing | A deal goes quiet for two weeks and the agent flags it for follow-up instead of letting it silently age out of anyone's attention until the opportunity is gone |
| Notion | Docs and Knowledge | A meeting produces a decision worth remembering and the agent captures and files it where the right person will actually find it later rather than letting it get lost in notes nobody revisits |
| Linear | Issue Tracking | An engineering issue sits unassigned for three days and the agent surfaces it in a daily summary rather than letting it get buried under a pile of newer tickets |
| Networking | A connection request goes unanswered and the agent handles a polite, appropriately timed follow-up rather than letting the opportunity quietly disappear because nobody remembered to circle back |
More agents are added over time based on what teams actually ask for. This list reflects where things stand today, not a fixed and permanent set of supported tools.
You Are Always the One Deciding
An agent exists in one of exactly two states at any given moment: running or paused. There is no ambiguous middle ground where you are not quite sure whether something is actively doing work on your behalf. That clarity is intentional, because ambiguity around what is acting in your name is exactly the kind of thing that erodes trust in automation over time.
Turn an agent on the moment you need it working. Pause it the instant you do not, whether that is because priorities shifted, something looks off, or you simply want a break from automation for a stretch of time. If you are finished with an agent entirely, delete it outright rather than leaving it dormant and half-forgotten, cluttering your dashboard with things you no longer use and no longer think about but that are technically still configured.
The decision about what runs on your behalf, when it runs, and when it stops is always yours. The agent handles the repetitive execution. You retain complete authority over whether it is running at all.
Nothing Happens Off the Record
Every single action an agent takes is logged without exception, regardless of how small or routine that action might be. A message sent, a record updated, a workflow triggered somewhere downstream, all of it lands in the activity log the moment it happens. This is not a feature we built as an afterthought to satisfy a compliance requirement. It is a core part of what makes it reasonable to trust an agent with real actions in the first place.
Knowing that everything is recorded means you can investigate anything. If something happens that you did not expect, you do not have to guess at what occurred or reconstruct a sequence of events from incomplete information. The log tells you exactly what happened and exactly when, in the order it happened, with nothing smoothed over or summarized away.
And because a log you have to remember to check is a log you will frequently not check, real time toast notifications surface meaningful activity the moment it occurs. You stay informed about what your agents are doing without having to actively monitor a dashboard. The information comes to you rather than waiting for you to go looking for it.
Getting Started
Getting an agent running is straightforward and does not require technical expertise or engineering involvement.
Open the marketplace and find the agent built for the tool your team already uses. Read the description to confirm it does what you need. Connect it to your account, a process that typically takes a couple of minutes and walks you through each step. Watch the live status indicator as it begins working so you can see immediately that it is active and functioning. Check the activity log whenever you want to review what has been done, and pause or remove the agent the moment it is no longer serving a clear purpose in your workflow.
That is the entire process from start to finish. No lengthy setup, no technical configuration, no ongoing maintenance required on your part.
Frequently asked questions
Quick answers to common questions.
Seven tools today: Greenhouse, Shopify, PagerDuty, HubSpot, Notion, Linear, and LinkedIn. More are actively planned and prioritized based on what teams using Kira actually request, rather than an internal roadmap developed in isolation from real user needs.
Because depth matters more than breadth for this kind of tool. Each agent is built to genuinely understand the tool it connects to and to handle real workflows rather than just firing off a generic action. We grow the marketplace deliberately and carefully rather than quickly and shallowly, because a shallow integration that fails when you rely on it is worse than no integration at all.
Completely, and by deliberate design. You run, pause, and delete agents entirely on your own terms without navigating complex menus or waiting on support. Every action they take is logged and confirmed as it happens so you always know what has been done on your behalf.
That depends on your plan. Starter plans support one active agent. Pro plans support up to ten. Business plans support up to twenty-five. If you need to discuss what is right for your team size and use case, reach out to support@kiraai.ai.
The activity log updates on every single action without exception and is available to review at any time. Toast notifications also confirm meaningful changes the moment they happen so you catch important activity in real time rather than discovering it after the fact.
Yes. Pausing stops the agent from actively running while keeping its full configuration completely intact. You can resume exactly where you left off without rebuilding or reconfiguring anything. Pausing is designed for situations where you want a temporary break from automation, not a permanent end to it.
If a third-party service experiences an outage or makes a change that affects the integration, the agent will stop taking action rather than behaving unpredictably. You will be notified so you are aware of the situation, and we work to restore full functionality as quickly as possible when the issue is on the third-party side.