Free forever, no credit card.Get Started for Free →
← All posts
November 19, 2026 · 6 min read

Does Greenhouse Remember Between Runs? The Candidate Record Persists. The Recruiting Work Does Not.

Does Greenhouse Remember Between Runs? The Candidate Record Persists. The Recruiting Work Does Not. Short answer: Greenhouse keeps the candidate record. It does not keep the recruiting work around that record. Applications, scorecards, and stage history stay attached to a profile. A scheduled agent still starts each run without what the last run concluded, unless that conclusion was written back into a field the next run knows to read. That split matters as hiring volume grows. The valuable pa

Does Greenhouse Remember Between Runs? The Candidate Record Persists. The Recruiting Work Does Not.

Short answer: Greenhouse keeps the candidate record. It does not keep the recruiting work around that record. Applications, scorecards, and stage history stay attached to a profile. A scheduled agent still starts each run without what the last run concluded, unless that conclusion was written back into a field the next run knows to read.

That split matters as hiring volume grows. The valuable part of a run is rarely the raw resume. It is the grouping, the exceptions, and the judgment about what needs attention next.

What the candidate profile actually keeps

The durable layer in Greenhouse is the person record. Every application and prospect record tied to a candidate stays linked to that candidate. Interview scorecards, stage timing, offer details, and activity history remain available from the profile.

That persistence is real. A recruiter opening a profile months later can see that the same person applied before, how far they progressed, and what interviewers wrote. A scheduled step can read the same fields again, because the data never left the system of record.

Notice what kind of memory that is. It is storage of submitted and entered data. It does not hold, on its own, a running account of what an automation noticed across several runs.

What Candidate Insights keeps, and where it stops

Candidate Insights is the clearest AI layer to test against this question. It answers natural-language questions about a single candidate, drawn from that candidate's Greenhouse data: application history, scorecard feedback, stage timing, and offer and salary details where the viewer has permission to see them.

Its scope is stated plainly in its own help pages. It is scoped to the individual candidate whose profile is open. It can reason across every application and prospect record tied to that person, but it cannot answer team-wide, aggregate, or cross-candidate questions. It is assistive only: it does not advance, reject, or take action on candidates, and it does not make hiring recommendations.

Its chat memory follows the same boundary. Conversations are saved per candidate profile, per job, and tied to the user account that started them. Another user does not see those chats, and a chat started on one candidate stays with that candidate. So even the conversation about a candidate persists only inside a narrow frame: one person, one job context, one viewer.

For a single lookup, that design works well. For an agent reviewing a whole pipeline overnight, it leaves the central question unanswered: what did yesterday's review learn, and what should today's check first?

What Talent Matching keeps

Talent Matching adds a second, narrower kind of persistence. It parses resumes and extracts skills, job titles, years of experience, employment dates, and company names, then compares that structured data against job criteria. Parsed resumes are cached for 30 days, and match results and reasoning are retained on the candidate at the point an application is rejected.

The pattern is consistent. The system remembers extracted fields and a match outcome attached to a record. It does not give a recurring agent a shared history of its own runs: which borderline cases it flagged last week, or which hiring manager asked for a second look.

Each feature remembers the object it was built around. None remembers the work that moves between objects, runs, and people.

What a scheduled run does not keep

Picture a common setup. A scheduled agent runs every morning over new applications for several roles. It groups duplicates, flags candidates who applied to two roles at once, notes where scorecard feedback disagrees, and prepares a short list of cases that need a human decision.

On the first morning, that grouping is the work product. The agent learns that two applications are the same person using different email addresses, that one role keeps attracting strong candidates from one source, and that three candidates are waiting only because a single interviewer has not submitted a scorecard.

When the run ends, the profile data is still there. The applications and scorecards can be read again tomorrow. The grouping, the source pattern, and the reason those three candidates were singled out are not stored as a shared history the next run can load. Unless a developer hand-builds a place for them and teaches every run to write and read it in exactly the same shape, the second morning re-derives the grouping, re-checks the scorecards, and re-builds the short list from the raw records again.

Multiply that across roles and weeks, and the cost is repeated reasoning. The record stayed perfect while the agent's learning evaporated between runs.

Give the runs one shared memory

Vilix AI is built for that missing layer: one memory that sits outside any single profile, job, or platform, and that every run can load before it starts and update before it ends.

It is cloud-hosted, so there is zero infrastructure to run or maintain. No extra database has to be provisioned, keyed to candidate IDs by hand, and kept in sync just so an agent can remember its own screening notes.

Every connected tool reaches the same memory over MCP. A screening agent beside Greenhouse, a drafting assistant preparing outreach, and a reporting step summarizing the pipeline can all read and write the same Vilix AI memory. The same memory follows the work everywhere, across tools and devices, instead of stopping at the boundary of one candidate profile or one user's chat history.

Vilix AI stores full conversation history, not just facts pulled out of it. For hiring work, that distinction is practical. A single extracted fact can say a candidate was flagged. The full history shows what the run saw, what it grouped together, and why it asked for a human decision, so the next run can verify the reasoning instead of repeating it.

With that layer in place, the morning routine changes. The first run saves its groupings, exceptions, and open decisions to Vilix AI. The second run loads that saved work first, then spends its time only on new applications and on the cases still waiting. Greenhouse keeps the record of record. Vilix AI keeps what the runs learned from working that record.

Start free, keep control of the data

Vilix AI has a free plan forever, with no time limit forcing a choice. Paid plans start with a 7-day Pro trial with no credit card required, so a hiring cycle can be tested before any payment.

Data stays under the team's control. Everything can be exported at any time, and everything can be deleted at any time.

Get Started for Free Greenhouse will keep doing its part well. Profiles will hold their applications and scorecards, and its AI features will keep answering questions inside the scope they were given. Add one shared memory for the scheduled work around those profiles, and the reasoning survives too. The next run then starts where the last run stopped, instead of reading the same records and calling it a fresh start.

Get Started for Free

Persistent memory across ChatGPT, Claude, and the AI tools you already use in Vilix AI.

Get Started for Free

Free forever, no credit card.

Keep reading
Does Braintree Remember Between Runs? The Vault Holds. The Run Reasoning Does Not.

Does Braintree Remember Between Runs? The Vault Holds. The Run Reasoning Does Not. Ask whether Braintree remembers between runs and the answer splits cleanly in two. Braintree remembers the payment state of a customer relationship in detail, because that is its job as a payment platform. It does not remember what an automation or an AI agent was thinking while it worked on that customer. Those are different kinds of memory, and mixing them up is where scheduled billing workflows quietly start r

Does Lemon Squeezy Remember Between Runs? The Customer Record Holds. The Run Reasoning Does Not.

Does Lemon Squeezy Remember Between Runs? The Customer Record Holds. The Run Reasoning Does Not. Ask whether Lemon Squeezy remembers between runs and the answer splits cleanly in two. Lemon Squeezy remembers the commercial state of a customer relationship in detail, because that is its job as a merchant of record for digital products. It does not remember what an automation or an AI agent was thinking while it worked on that customer. Those are different kinds of memory, and mixing them up is w

Does FastSpring Remember Between Runs? The Account Record Holds. The Run Reasoning Does Not.

Does FastSpring Remember Between Runs? The Account Record Holds. The Run Reasoning Does Not. Ask whether FastSpring remembers between runs and the answer splits cleanly in two. FastSpring remembers the commercial state of a customer relationship in detail, because that is its job as a merchant of record. It does not remember what an automation or an AI agent was thinking while it worked on that customer. Those are different kinds of memory, and mixing them up is where scheduled billing workflow