Ask what's waiting on you. Clear it from Claude Code or ChatGPT.
Your assistant works the same records your team does: candidates, scorecards, security reports, and salary bands, through 200+ MCP tools.
Start with the built-in assistant in the browser, or connect your own over MCP. Each of the 200+ tools declares whether it reads data, changes a record, or emails a real person, before the client runs it.
ChatGPT can draft the job ad. It can't see who applied.
A useful assistant needs access to the records behind its answer.
Give the assistant the context it needs.
Interview preparation needs the candidate's application and prior reviews. Salary research needs your band and current listings.
Know where candidate data goes.
Before connecting candidate data, check which provider receives it, how it is used, and whether you can change models.
Stop copying records into a chat.
Without access to your records, an assistant needs you to paste in resumes, reviews, and salary bands each time you ask.
Which reviews are waiting on me this week?
Ask a question about your hiring pipeline.
Ask about candidates, stages, reviews, and salary bands. The conversation shows which tools the agent called.
Try a question
Agents for hiring, security, outreach, and training. 200+ MCP tools.
Each agent has its own tools and instructions. What it can reach depends on your account permissions.
Hiring Agent
Read candidate records, compare salary bands, draft messages, and send interview invitations. Tool calls appear in the conversation.
Security Agent
Review vulnerability reports, compare severity and bounty amounts, and draft replies to researchers.
Outreach Agent
Check campaign results, add prospects, and draft emails and replies. Available with the Outreach add-on.
Training Agent
Create a program from a template, edit slides and quizzes, invite people, and check who has finished.
@KitBot in Slack
Mention @KitBot in a candidate channel to read the record, save a note, or request a stage change. It also answers in your hiring and security channels.
Choose your AI provider. Switch anytime.
Connect Gemini, Anthropic, or OpenRouter. A stronger model handles analysis and a cheaper one handles routine replies; with OpenRouter you pick each model yourself. Your provider bills you directly, with no Kit markup on tokens.
- Use the included AI allowance, or bring your own API key
- Pay your provider directly, with no markup on tokens
- A stronger or cheaper model per task; OpenRouter lets you choose each one
- Daily and monthly spending limits on your own key
- Usage and cost by team member and by model
Run the agents on the included AI allowance or your own key.
Free for 30 days. Card required. Cancel anytime.
200+ tools for your existing assistant.
Connect Claude Desktop, ChatGPT, Cursor, or Copilot to your Kit records. Tools declare their effects before a client runs them. Access depends on your permissions and approved scopes.
- Sign in with OAuth, no API key to paste
- You approve each connection and choose which account it can reach
- Each client gets only the scopes you approve
- Search docs, resumes, and reports by meaning
- Public read-only endpoint for docs and templates
OAuth 2.0 with PKCE. Scoped tokens. EU infrastructure, signed DPA on request.
Work that runs in the background.
In the background, AI screens security reports, sorts inbound email, extracts resume fields, drafts outreach, and classifies replies.
Inbound Email Triage
Matches catch-all emails to open roles. It can create an application, ask for missing details, or send a decline.
Outreach Email Drafting
Reads the prospect's company, role, and recent work, then drafts an email using your campaign's structure. You review it before sending.
Reply Analysis
Classifies the reply, reads the conversation, and drafts a follow-up for you to review and approve.
Security Report Screening
Checks incoming reports for spam signals, including invented CVEs and code paths. Flagged reports carry a score and the reasons for it.
Resume Data Extraction
Extracts your custom fields from resumes, forms, and questionnaires, with a confidence score for each value. Results can appear in Slack candidate channels.
What you can configure in Kit.
Check tool access, model choice, usage limits, and data permissions before connecting an assistant to your hiring records.
| Capability | Kit | What to check elsewhere |
|---|---|---|
| AI model choice | BYOK: Gemini, Anthropic, OpenRouter | Can you choose and change the model? |
| MCP tool access | 200+ tools for Claude, ChatGPT, Cursor | Can your existing client connect? |
| AI agents | Agents for hiring, security, outreach, and training | Which records and actions can it use? |
| Data governance | Account permissions and scoped tool access | Who can access which records? |
| Cost visibility | Spending limits and a usage dashboard | Can you see and limit model spending? |
| AI pricing | Included AI allowance, or your own provider key | Are usage charges separate from seats? |
| Autonomous workflows | Report screening, email sorting, extraction, and drafting | Which tasks can run in the background? |
| Output attribution | Model and usage records | Can you trace how an output was produced? |
Tools connected to the work.
Read records, draft messages, and extract fields through agents and 200+ MCP tools.
Agents for specific tasks
The Hiring, Security, Outreach, and Training agents and @KitBot in Slack each have their own instructions and tools.
Spending limits
Every seat includes an AI allowance. With your own key, set daily or monthly spending limits per provider. Admins get a warning before a limit is reached, and the usage dashboard shows cost by team member and by model.
A model per task
A stronger model for analysis, a cheaper one for routine replies. With OpenRouter you choose each model yourself.
Structured Outputs
Extract candidate fields, compare salary bands, and record severity ratings as structured data you can review.
Model controls
Choose your provider, inspect usage records, and decide which AI features you use. Your team remains responsible for hiring decisions.
Works Without AI
Manage candidates, stages, and interviews even when your model provider is unavailable. Most drafts and summaries catch up once the provider is back.
Start with a question about a candidate.
Try the Hiring Agent in the browser. You can connect an external assistant later.
Start on the included allowance or your own key
Use the included AI allowance, or add a Gemini, Anthropic, or OpenRouter key in Settings.
Set spending limits
With your own key, set daily and monthly limits for each provider. You can change them later.
Ask about an application
Open the Hiring Agent in Kit and ask for a candidate summary. To connect an external assistant, copy Kit's MCP server URL from Integrations.
One key for everything in Kit.
Use the same provider key and spending limits across hiring, security, training, outreach, and Compensation Research (beta). Everything except Outreach is included in every seat; Outreach is a paid add-on.
Questions about AI in Kit.
What is Kit?
One account for hiring, security reports, security training, and Compensation Research (beta). Work in the browser, use Kit's agents on the included AI allowance, or connect an assistant through 200+ MCP tools. Outreach is the only separately billed add-on.
What providers and models does Kit support?
Google Gemini, Anthropic, and OpenRouter. A stronger or cheaper model is picked for each kind of work; with OpenRouter you choose each model yourself. Search embeddings always run on Gemini. You can change providers in Settings.
Do I need my own API key?
No. Every seat comes with an included AI allowance that resets daily and monthly, so you can try the agents first. Connect your own Gemini, Anthropic, or OpenRouter key when you need more, and set your own spending limits.
What data does the AI access?
Agents access records within your Kit account, subject to permissions. MCP clients use scoped OAuth tokens, so access depends on the scopes you approve. Connected model providers process the data needed for those requests under their own terms.
Does using Kit establish EU AI Act compliance?
You choose the provider, inspect usage records, and decide which AI features you use. Those controls do not establish EU AI Act compliance on their own. Your obligations depend on how you use the models in hiring.
How much does AI cost?
An AI allowance is included in every seat. With your own key, your provider bills you directly with no Kit markup, and you set daily and monthly spending limits.
What is MCP?
MCP (Model Context Protocol) is an open standard that lets clients like Claude Desktop, ChatGPT, and Cursor connect to external services. Kit exposes more than 200 MCP tools across hiring, security, training, outreach, and compensation. Connect Kit as an MCP server and your client can list candidates, advance applications, and check salary benchmarks.
Can I use Kit without AI?
Yes. Candidates, stages, reviews, and interviews all work without AI. Leave AI-extracted fields off and skip the assistant; drafts, summaries, and extraction then don't run.
What happens if a provider goes down?
Candidates, stages, and interviews keep working. AI drafts and summaries wait and resume once the provider is back. You can switch your preferred provider in Settings at any time.
Start in Kit. Bring your assistant when you're ready.
Use the included AI allowance or connect your own key. Free for 30 days. Card required. Cancel anytime.