PlanS
Storage25 GBfiles and history
$199per month$1,999per year, paid upfront
Pitstop Executive search
AI agents for executive search, with a partner’s yes on every message
AI agents can prepare six of the stages of a search: market map, longlist, profiles, approaches, interview notes and the client update. A person approves anything that reaches a candidate or a client.
Where AI fits in your searchOne search firm, three months
- 755executives contacted by the agent
- 24%replied
- 139qualified meetings booked
- 90meetings held
Where AI fits in your search, stage by stage
Tick the stages your firm runs and type your own numbers. The board tells you how many messages per mandate will wait for a partner’s yes, which level to set, and where to start.
| We do this | Stage | What an agent can prepare | What it may do alone | Who decides |
|---|---|---|---|---|
| The target companies and the people in matching roles, with the public sources it used. | Build the map internally. | The consultant reviews the map. | ||
| A ranked list against the brief, with the reason for each name. | Rank internally, as a proposal. | The consultant. Under the EU AI Act, evaluating candidates is a high-risk use. | ||
| A one-page note from the CV and the public record, with sources and gaps. | Draft internally. | The consultant, before it goes to the client. | ||
| A short, personal message per candidate, then the next touch with what changed. | Nothing: it asks first. | The consultant or partner approves the exact text. | ||
| A summary and a draft scorecard against the brief. | Draft internally. | A person, before it reaches the client. | ||
| The weekly shortlist note. | Nothing: it asks first. | The engagement partner, or the client through a one-time link. |
Your numbers. The two timings are assumptions: replace them with yours.
Pitstop does not do the research itself: your agents do (ChatGPT, Claude, Codex or ours), and Pitstop sets what they may do alone.
What can go wrong, and what the law asks
One message can lose a mandate.
A candidate approached with the wrong company name. A confidential search mentioned to someone who knows the incumbent. A client update sent before the partner saw it. So most firms keep AI on drafts and send everything by hand: the agent saves an hour of writing and costs an hour of copying.
The first approach carries a legal duty.
In Europe, when you contact a candidate whose details you found elsewhere, GDPR Article 14 requires telling them who you are, why you hold their data and what their rights are, at the latest in that first message. An approach that leaves unread can miss it.
One mandate, and where the partner says yes
One approach on the sample mandate, from draft to journal
- Sourcing agentDrafts the first approach to Claire M., with the Article 14 notice line.
- Pit stopIt waits at L2. The consultant reads the exact text.
- Partner’s yesApproved and fingerprinted (1f4a…f1d7). Any later edit voids it.
- SentThe agent sends and reports what went out.
- JournalPitstop checks the sent text against the yes and records who, what, when.
CFO search · sample mandate
Sample mandate. Approve the first approach, then edit it: the fingerprint is computed in your browser.
| Stage | Agent | Who decides | Status |
|---|---|---|---|
| LonglistSearches and ranks profiles against the brief. | Scout | Internal only: it may act, then report | Runs alone |
First approachDrafts a short, personal message per candidate. | Sourcing | Waits for the consultant’s or partner’s yes | Awaiting partner |
Content fingerprint | |||
| Follow-upsPrepares the second and third touch, with what changed. | Sourcing | Waits for a yes, every time | Hold |
| Client updateDrafts the weekly shortlist update. | Memo | Waits for the engagement partner | Awaiting partner |
| Interview notesTurns notes into a summary for the client. | Minutes | Waits for a yes before it leaves | Awaiting partner |
Each approval is tied to the exact text. If someone edits the approach after the partner said yes, the approval no longer counts and the agent asks again.
The AI tools search firms use, and where Pitstop fits
Pitstop is not a sourcing tool or an ATS. It is the layer where the agents you already use ask before anything leaves.
| Job | Examples | What they do | With Pitstop |
|---|---|---|---|
| Sourcing | hireEZ, SeekOut | Search profiles across many platforms and rank them against a role; SeekOut works inside the AI assistant your firm already approved. | Keep them for finding names. The approach the agent drafts stops at a pit stop. |
| ATS and search CRM with AI | Bullhorn Amplify, Recruiterflow | AI inside the system of record: screening and matching (Bullhorn Amplify), call notes and sourcing (Recruiterflow’s AIRA). | Keep your ATS as the record. Pitstop holds the approvals and the journal around it. |
| Interview notes | Metaview, Recruiterflow’s notetaker | Turn interviews and calls into notes and summaries. | Notes stay internal. The summary sent to the client waits for a yes. |
| Research and drafting | ChatGPT, Claude, Codex | General assistants that map markets, write profiles and draft messages, and now connect to mailboxes. | They connect to Pitstop once and ask before they send. |
Product pages checked on 11 October 2026. Names are examples, not a ranking; we have no commercial tie with these vendors.
One morning on a search, and what running it taught us
AI Jungle runs agents every day for firms in executive search and business development. Each morning they rank who is worth contacting, draft the first message and prepare follow-ups, and a partner decides. What we learned is in Pitstop:
Overnight, the sourcing agent ranked the longlist against the brief. Internal work: no yes needed.
internalThree first approaches and two follow-ups wait in Validations, each with its fingerprint.
waitingThe partner approves three, edits one line in a fourth and refuses one, with a reason for the agent.
partner’s yesThe agent sends what was approved and reports it. Pitstop checks each send against the approved text.
checked against the yesThe client asks who approved Tuesday’s shortlist note. The partner opens the journal.
journal
An illustrative morning, not a client case.
- A short message beats a clever one. Our agents write to one person, about one thing that only concerns them, with one question. Pitstop is where a partner checks that before it goes.
- Approval has to be fast. One inbox, on laptop or phone. A refusal needs a reason, and the reason goes back to the agent.
- Confidential means traceable. Who approved which message to which candidate, kept in a journal you can verify and export.
Start with one mandate, for 30 days
- Pick one live mandate. And the consultant who will approve every approach.
- Connect the agent you already use. ChatGPT, Claude or Codex, once, with OAuth. It drafts the longlist notes and the first approaches.
- Keep every send at L2, ask first. The consultant approves the exact text; refusals go back to the agent with a reason.
- After 30 days, compare. Replies and meetings against your previous mandate, and how many approaches were approved as is, edited or refused.
Free for one person with the plugin; for the firm, one plan below. Prefer to have the agents built and run for you? AI Jungle’s agency starts with a Diagnostic. Diagnostic.
Pricing for search firms
One price for the firm, never per consultant: a 4-person and a 40-person search firm pay the same plan. Storage fills with files, not people. Most firms start on S; choose M or L if you keep many CVs, call recordings or years of mandate history in Pitstop. Choose a plan and we set up your workspace with you within 2 business days. Compare the plans in detail.
Plans per firm, in USD · Whole firm
Every plan includes
- The nine workspace areas, for the whole firm
- Pit stops, autonomy levels and a journal where any edit shows
- One-time approval links for clients
- Roles and invitations
- Hosting in Germany (EU)
Choose a plan and we email you within 2 business days to set up your firm’s workspace with you, then we switch your plan on.
Questions search firms ask
Candidates and clients
Will candidates know an agent drafted the message?
That is your firm’s policy. Pitstop makes sure a person approved the exact text before it went out.
Can a client approve our shortlist update directly?
Yes, with a one-time approval link, no account needed.
Can a partner refuse a message?
Yes. A refusal needs a reason, which goes back to the agent with the request.
Tools and setup
Do our consultants need to learn a new tool?
No. Your team keeps using ChatGPT, Claude or Codex. The agents connect to Pitstop once, with OAuth, and ask whenever something is about to leave the firm. Decisions happen in one inbox, or through a one-time link for a client.
Our ATS is Bullhorn, Invenias or Ezekia. Does Pitstop connect?
Pitstop sits between your agents and the outside world, not inside your ATS. Your agents keep their own access.
Do you also build the agents?
AI Jungle’s agency does, for search firms specifically, starting with a Diagnostic.
Where is our data?
On AI Jungle’s own servers in Germany (EU).
Why not let ChatGPT or Claude send from our mailbox directly?
They can now connect to mailboxes and calendars. What they do not keep is a record of which exact text a partner approved before it left. With Pitstop the agent asks, a partner approves the exact text, and the journal shows any gap.
Is ranking candidates with AI regulated in Europe?
The EU AI Act lists AI used to evaluate candidates among high-risk uses (Annex III, point 4(a)), with human-oversight duties (Article 14). Treat the agent’s longlist as a proposal a consultant reviews, and keep the record.
Does the first approach have to say where we found the candidate?
In Europe, GDPR Article 14 requires telling a candidate who you are, why you hold their data and what their rights are, at the latest at the first communication. One more reason a person should read the first approach before it goes.
Let your agents do the outreach work. Keep the last word.
One price for the firm, never per consultant. We set up your workspace with you within 2 business days.