> ## Documentation Index
> Fetch the complete documentation index at: https://guide.kello.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Kello Alpha usage, limits, connected apps and trust

> Kello Alpha has a usage budget per user, connects to Apollo, Gmail and more with your approval, and says what it can't do and what it offers instead.

Kello Alpha has a usage budget per user, measured by the work it does for you, not a count of searches. Kello's own data has no emails or phone numbers, but you can connect your own tools, such as Apollo, Gmail, FullEnrich and Granola, and Kello Alpha uses them with your approval on every send or write.

> [card] [Usage and budget](#what-uses-my-budget)

> [card] [Connect an app](#how-do-i-connect-an-app)

> [card] [What Alpha can't do](#quick-reference)

> [card] [Why Alpha refused a search](#5-no-discrimination-by-design)

> [card] [Gaps in the data](#is-anything-thin-in-the-data)

## What uses my budget?

The work Alpha does on your messages: reading what you sent, searching, checking its results and writing the answer. It's not a fixed price per search. A one-person lookup uses little. A long, open brief with several rounds of checking uses the most.

## How long does an answer take?

A first search takes about 3 minutes; each follow-up takes about 1. Very open briefs ("think creatively …") take longest. Break them into a clear first ask plus short follow-ups. See [get better results](https://guide.kello.ai/alpha/better-results.md#start-broad-then-narrow).

For timings by kind of ask, see [how long does each kind of ask take?](https://guide.kello.ai/alpha/better-results.md#how-long-does-each-kind-of-ask-take)

You can leave the page while Alpha works. The answer keeps running and is there when you come back. **Stop** really stops it and keeps the partial answer.

## Where do I see my usage?

Go to **My Account** and open **Usage** (shown as **Plan and usage** on some accounts). You'll see your search budget as a percentage remaining.

## Will I be warned before I run out?

Yes. Kello emails you when you've used 50%, 80% and 90% of your budget.

## What happens when I run out?

The answer stops with a message that points you to the Kello team. Contact them and they'll sort out more budget.

A run can stop partway if the budget runs out mid-answer. When you're close to your limit, ask narrow questions: one company, one place, one follow-up at a time.

> [!TIP]
>
> If you send a message and Alpha doesn't reply at all, check your usage under My Account. An empty budget is the most common reason.

## What plans are there?

* **Free accounts** get a small monthly allowance.
* **Paid plans** are set up by the Kello team and shared across your team.

There's no public price list or self-serve checkout. To get more, or to set up a plan for your team, email [support@kello.ai](mailto:support@kello.ai) or [talk to a person](https://calendly.com/mona-gandhi-kello/ama).

## How do I make my budget go further?

* **Be specific up front.** A clear first ask saves rounds. See [get better results](https://guide.kello.ai/alpha/better-results.md).
* **Refine the list you have** instead of starting new searches from scratch.
* **Don't resend while Alpha is working.** Come back and the answer is there.
* **Use Stop** if Alpha is heading the wrong way, then correct course.

## Is Kello (done-for-you) part of my Alpha budget?

No. Kello, the done-for-you service, is separate from Alpha usage. See the [Kello quickstart](https://guide.kello.ai/kello/quickstart.md).

***

## Which apps can I connect?

Teams have connected **Apollo**, **Gmail**, **FullEnrich** and **Granola** from chat, and an **ATS (Applicant tracking system: the software your team uses to track candidates.)** through a custom connection. Use them for work emails, email drafts and meeting notes, alongside Alpha's search.

## How do I connect an app?

Ask in chat.

> [!WARNING]
>
> **Apollo needs a master API key (A password-like code that lets one app connect to another.)** for person lookups. A normal key connects, but every lookup then fails. Create a master key in Apollo before you connect.

A typical flow:

**1. Ask to connect**

```text Ask to connect wrap theme={null}
can i connect my apollo.io account here?
```

Alpha sends a secure connect link.

**2. Say you've connected**

```text Confirm the connection wrap theme={null}
i have connected Apollo
```

Alpha confirms it works.

**3. Ask for work emails**

```text Find work emails wrap theme={null}
use apollo to find their work emails
```

Work emails for the people on your list.

**4. Ask for a draft**

```text Draft an email wrap theme={null}
can you draft an email to reach out to [name]
```

A draft in chat.

**5. Push it to Gmail**

```text Save as a Gmail draft wrap theme={null}
push this as draft on my email account
```

Saved as a Gmail draft, after you approve.

> [!WARNING]
>
> Never paste an API key into chat. Alpha refuses it and sends you the secure connect link instead.

* **Work emails are fine; personal emails and phone numbers are refused**, even through a connected tool.
* **Anything that sends or writes waits for your approval.** You see what will happen first, and can reject it.

For an ATS, talk to the Kello team at [support@kello.ai](mailto:support@kello.ai). It's set up as a custom connection.

***

## What can't Alpha do, and what does it do instead?

What Kello Alpha can't do, and what it offers instead.

| You ask for | What Alpha does instead |
| - | - |
| People by age, gender, caste, religion or ethnicity | Refuses. Offers a fair stand-in where one exists: "VCs in the Bay Area who studied in India" (63 people) instead of "VCs of Indian origin". |
| A salary band ("current CTC (Cost to company: the total yearly pay package.) 60–70 LPA (Lakhs per annum: yearly pay in Indian rupees (1 lakh = 100,000).)") | Turns it into a seniority and experience range, and says so |
| Notice period (How long someone must keep working at their current job after resigning.), visa status or GPA | Uses "open to work" as the nearest signal |
| Emails or phone numbers from Kello's data | None stored. Use a connected tool for work emails. |
| Company events: acquired, filed for IPO, raised last week | Uses the web, or a stand-in such as "led legal through an IPO", and says which |
| Day-level timing ("quit in the last 7 days") | Job dates are by month, so you get the most recent movers |
| Adding people to a collection from chat | Add them from the results list |
| Several lists in one message | Builds the first and offers the next. Ask for one list per message. |
| File uploads | Paste the text, or a link |
| Memory of past conversations | Keep refining in one chat, or restate key points in a new one |
| Sort, filter or bulk select on the results page | Ask Alpha in chat |
| An export button | Ask `export this as CSV` |
| Help applying for a job | Says Kello is for hiring teams and points to the company's careers page |
| Anything that isn't recruiting | One polite line, then an offer to help with hiring |

> [accordion] Can I get contact details?
>
> Not from Kello's own data, which has no emails or phone numbers. Connect a tool such as Apollo or FullEnrich and ask Alpha to find work emails for the people you choose (see [how to connect](#how-do-i-connect-an-app)). Personal emails and phone numbers are refused. If you want someone to reach out for you, send the role to [Kello](https://guide.kello.ai/kello/quickstart.md).

> [accordion] Can Alpha tell me what someone earns?
>
> No. Salary is rarely in profiles, so Alpha can't benchmark pay or filter on it. If you give a pay band, Alpha turns it into a seniority and experience range and tells you so.

> [accordion] Can I filter on notice period, visa or GPA?
>
> No. The nearest signal is "open to work", which Alpha can filter on. Check notice period and visa in your first conversation.

> [accordion] Can I upload a JD or résumé?
>
> Not as a file. Paste the text, or a link to the job post. Long pastes are fine.

> [accordion] Does Alpha search my applicants or ATS?
>
> Alpha's search runs on Kello's people data and your own Kello collections, not your applicants. Save people to a collection, then ask `search my [collection name] collection for …`. Only members with a Kello profile can be searched, so applicants who aren't in Kello's people data won't be found. Alpha tells you how many members it can search. To connect an ATS, talk to the Kello team.

> [accordion] Can Alpha send messages for me?
>
> Alpha doesn't run outreach sequences. With Gmail connected, it can draft an email and save it as a Gmail draft. Every send or write waits for your approval.

> [accordion] Does Alpha remember my earlier chats?
>
> No. Each conversation stands alone. In a new chat, restate what matters: `no agency recruiters, product companies only`.

> [accordion] Can I rename, delete or share a conversation?
>
> Not today. Share a candidate's public link, a collection, or a CSV (A spreadsheet file that opens in Excel or Google Sheets.) instead.

> [accordion] Can I sort or filter the results page?
>
> Not on the page itself. Ask in chat: `only Mumbai`, `drop anyone at a services company`.

> [accordion] How do I see the rest of a long list?
>
> Open the full list: the chat shows only a preview. The results page shows 100 people per page. See [how do I see the whole list?](https://guide.kello.ai/alpha/working-with-results.md#how-do-i-see-the-whole-list)

> [accordion] Where is the export button?
>
> There isn't one yet. Ask `export this as CSV` in chat.

***

## How does Alpha keep results honest?

Alpha follows eight principles. It never makes up facts, tells you what it changed, checks its own work and stays on task.

### 1. It never makes up facts

Lists come from real records. Facts from the web are cited. If Alpha doesn't know something, it says so.

### 2. It tells you what it changed

If a must-have became a nice-to-have to find enough people, Alpha says so. If something can't be filtered on, it says that in plain words and offers the nearest substitute. Read the last sentence of each answer: that's where Alpha says what it assumed.

### 3. It checks its own work

Before you see any people search, a second review checks the top results against the person you described: what fits, what leaked in, what's missing. Alpha tightens or widens the search, at most twice, then shows you the list.

### 4. Your examples are respected

When you paste LinkedIn profiles as examples in the same message as your ask, a search that would reject any one of them cannot be saved. See [how do I add examples?](https://guide.kello.ai/alpha/finding-people.md#how-do-i-add-examples)

### 5. No discrimination by design

Alpha refuses searches on religion, caste, ethnicity, gender, sexual orientation, disability, marital status or age. Where a fair substitute exists, it offers one, such as an experience band instead of an age, or "studied or worked in India" instead of "of Indian origin". Searches about a person's private life are declined.

### 6. No contact scraping

Kello's data never exposes emails or phone numbers, not in search, lists or exports. Through a connected tool, Alpha fetches work emails only. Personal emails and phone numbers are refused.

### 7. Nothing happens in your apps without your click

When you connect your own tools, every send or write shows a **Needs your approval** card with the app, the action and a preview. Nothing happens without your click. **Reject** skips only that action.

### 8. It stays on task

Alpha does recruiting and talent work only. It ignores instructions hidden inside profiles, job descriptions or web pages, so a pasted document can't redirect it.

***

## What data does Alpha search?

Alpha searches Kello's data on 830M+ professionals and 57M companies, built from public professional information, plus live job postings and your own Kello collections when you ask. See the [privacy policy](https://kello.ai/privacy-policy).

## Is anything thin in the data?

Work history is rich: dates, titles, seniority, tenure, and company stage and size at the time. Some gaps change how you should phrase an ask:

* **Role descriptions are thin.** Only about a third of past roles have a written description. Asks about what someone did inside a job ("owned retention curves") find fewer people than really exist. Pair them with titles or companies.
* **Profiles lag.** New jobs show up weeks to months late, so someone who moved recently may still show their old employer. For one person, a LinkedIn URL is the surest way in.
* **Counts stop at "10,000+".** The review looks at the top of the list, so with a huge pool the ranking does the work. If the top looks off, narrow the pool.
* **Niche specialties may come back empty.** "Astrodynamics experts in India" found mostly the wrong people. Name the places these people work ("ISRO, Digantara, Skyroot, Pixxel") or their exact titles ("flight dynamics engineer") so Alpha has something firm to search on.
* **Skills are sparse.** Skills are listed on only part of profiles. Describe the work and career, not only skill keywords.

## Related

> [card] [Better results](https://guide.kello.ai/alpha/better-results.md)
>
> Phrase asks around the gaps.

> [card] [FAQ](https://guide.kello.ai/alpha/faq.md)
>
> Quick answers, including why Alpha refused a search.

> [card] [Working with results](https://guide.kello.ai/alpha/working-with-results.md)
>
> Collections, removals and CSV export.

> [card] [Quickstart](https://guide.kello.ai/kello/quickstart.md)
>
> The done-for-you service.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.