Kepler vs Spott: which AI-native recruitment CRM should an agency choose?

What Kepler and Spott each are

Spott describes itself as an AI-native ATS and CRM for recruitment agencies. It is a broad, fast-moving product: its changelog runs weekly and its documentation covers candidates, companies, contacts, opportunities, jobs, placements, a unified inbox, calls, calendar, notes, tasks, lists, outreach campaigns, reporting, and roughly thirty integrations. Its approach to AI is to ship many discrete features — CV diffing, enrichment, AI-filled columns, CSV mapping, matching, text-to-filters, inbound triage, call mining, job-description generation, campaign copy, scorecards, CV reformatting — each with its own settings toggle and, in most cases, its own credit cost.

Kepler is a recruitment CRM built around a single assistant and a searchable database. Rather than twenty AI touchpoints, it has one agent that sits in a sidebar on every page, reads the whole workspace — candidates, CVs, notes, emails and call transcripts — and can create and update records once a person approves the change. Around that sit the ordinary parts of an agency desk: pipelines, jobs, tasks, email and calendar sync, a meeting notetaker, CV parsing and formatting, a branded job board, reporting, campaigns and automations.

Both products are new relative to the incumbents, and both are competing against Bullhorn, Vincere and JobAdder rather than against each other in most deals. But when they do meet, the choice is unusually clear-cut, because they disagree about something fundamental.

The real difference: where the AI lives

Spott’s Ask AI is a panel scoped to a record. You open a candidate, a job, an application, an opportunity or a call, and you can ask questions about that record. For work that spans many records — “which of my Berlin candidates has actually shipped a payments integration, and who owns them?” — Spott’s own documentation points you at their MCP server, which exposes your Spott data to ChatGPT or Claude so an external assistant can query it.

That is a defensible design. It is cheap to build, it keeps the product’s surface area small, and it lets Spott ride whatever frontier model the customer already pays for. The cost is that the cross-record work happens somewhere your consultants do not work, with no write path back into the CRM, and no shared context with the rest of the team.

Kepler makes the opposite bet. The assistant is inside the product, on every page, with roughly fifteen tools available to it. It can search the entire database semantically and by keyword, pull a shortlist for a job, save that shortlist as a list, and create or update records — with an approval step before anything is written. Because it is in the product, its results are records rather than text: a returned set of candidates is a set you can act on, not a list of names you retype.

The second-order consequence is what search reads. Spott’s AI-ranked search covers notes and attachments labelled as CVs. Kepler’s covers those plus the email threads and call transcripts attached to a record, indexed as text and retrieved with a hybrid of semantic and keyword search. For an agency whose real institutional memory is eight years of email and a folder of transcripts, that difference is the product.

How each one matches candidates to jobs

The user experience is similar enough that a demo will not separate them. Both let you write requirements for a job in natural language, both pre-filter the pool, and both return a ranked list with a per-criterion breakdown. Spott shows a 0–100 score with a thumb up or down per criterion. Kepler shows a strength label with a score, and expands under the row into the evidence: for every requirement, a verdict and the text from the candidate’s record that supports it.

Two differences matter in practice. The first is auditability. A consultant who disagrees with a ranking needs to see why the system ranked that way, and a thumb does not carry a reason. The second is measurement. Kepler runs a standing evaluation programme over its matching, with recall and precision gates and a hard cost ceiling per job, so a change to the ranker has to prove itself before it ships. Spott publishes no equivalent, and their documentation notes candidly that matching quality degrades when the hard-filter pool exceeds roughly two thousand candidates.

That last point deserves credit rather than a jab: Spott documents its own sharp edges unusually well, and an agency comparing the two should read their documentation directly rather than taking either vendor’s summary of it.

Where Spott is the better product

This section is the honest half, and it is not short. On breadth of agency workflow, Spott is ahead of Kepler today in at least eight places.

• Outbound sequencing. Kepler has campaigns — multi-step email sequences with audiences, delivery windows, exit criteria and reply classification. Spott’s are further along: manual steps that materialise as assignable tasks, per-mailbox daily send caps with randomised jitter, automatic warm-up for new mailboxes, delegated sending, and campaign templates. If sequencing is the centre of your desk, this gap is decisive.
• Opportunities. Spott models pre-job business development as its own object, with stages that each carry a closing-rate percentage, so a board shows weighted pipeline value and an opportunity converts into a job. Kepler has no opportunities object and no weighted forecasting.
• Interview scorecards. Spott generates a scorecard template on the job and captures per-application scores and comments against it. Kepler has none.
• A client-facing portal. Spott lets you invite a client contact to review a shortlist, with per-field visibility controls, comments, ratings and interview requests. Kepler publishes a job board but has no client review portal.
• Consent and GDPR as a data model. Spott stores consent records with purpose, lawful basis, source and validity period, and can filter on expiring consent. Kepler has a privacy posture but no consent object, which matters for EU agencies with a compliance officer.
• Telephony. Spott integrates seven VoIP providers with click-to-call, inbound call cards, recording fetch and transcription. Kepler records meetings but integrates no diallers.
• Multiposting. Spott distributes jobs through Broadbean and idibu and links LinkedIn Recruiter projects. Kepler publishes to its own branded job board and nowhere else.
• A mobile app. Spott ships iOS and Android, including in-person meeting recording. Kepler’s web app is responsive but there is no native app.

Spott also holds ISO 27001 certification and hosts in the EU. Kepler holds no security certification at present. For an agency whose clients run vendor security reviews, that is a straightforward reason to choose Spott, and no amount of product depth substitutes for it.

Where Kepler is the better product

Four areas, and they cluster around trusting what is in the database rather than adding another module to it.

• The assistant’s reach. Covered above: whole-workspace, inside the product, able to write back after approval.
• Provenance and reversibility. Kepler keeps history at the level of the individual field, records the source of every AI-suggested change, and supports undo with conflict detection. Spott’s activity panel shows that a change happened and where it came from, but there is no per-cell history and no undo. Relatedly, Kepler uses optimistic concurrency across its resources, so when two consultants edit the same record the stale write is refused rather than silently applied — an unglamorous property that quietly prevents a class of data loss.
• Proposed rather than applied. Both products mine conversations to keep profiles current. Kepler routes every derived change into a proposal a person accepts or rejects, with the email, call or message that produced it attached. Spott’s scheduled profile refresh documents no review surface and its history reports only whether a profile was updated or not found, which implies changes are applied directly. For a database a team has manually corrected over years, the review step is worth having.
• The careers site. Kepler publishes jobs to your own domain, server-rendered and indexable, with structured data per role. Spott hosts branded career pages on a path under their own domain. If you want your roles to accumulate search authority for your brand rather than the vendor’s, that distinction compounds.

How much does each one cost?

Kepler publishes one price: USD 119 per user per month billed annually, which is USD 1,428 per seat per year. There is one edition and every feature is included — there is no tier above it and no feature gating. Records and storage are unlimited, with no per-record or per-contact charge. Migration is free on every plan. Additional AI capacity is available on request; a monthly AI allowance is included without it.

Spott does not publish per-seat pricing on its website. Its documentation describes two seat plans, CORE and PRO, alongside a separate pooled credit balance used for AI and enrichment actions. Credit costs are published per action — a work email costs 1 credit, a personal email 3, a phone number 10, a LinkedIn profile refresh 5 — and credits carry over and expire after a year. To compare totals you would need a quote from Spott and an estimate of your own enrichment volume.

Neither model is obviously better. Spott’s credit system lets genuinely expensive AI features ship without eroding their margin, and an agency that uses little enrichment may pay less overall. Kepler’s single price means there is nothing to forecast and nothing to gate, at the cost of less granular control. What is fair to say is that the two require different amounts of work to evaluate.

Which one is right for your agency?

Choose Spott if outbound sequencing is how your desk generates revenue; if you need a business-development pipeline with weighted values; if your clients expect a portal where they review and comment on a shortlist; if you need interview scorecards, a dialler integration, job multiposting or a mobile app; if you operate in the EU with a compliance function that wants consent modelled as data; or if your clients’ security reviews require ISO 27001. Spott ships all of that now and Kepler does not.

Choose Kepler if your main asset is a large database you cannot currently search — years of CVs, notes, email threads and call transcripts that no keyword search reaches; if you want one assistant that answers a real question across all of it and can then act on the answer; if you need to audit why a candidate ranked where they did; if you care that automatic profile updates are proposed rather than silently applied; or if you want one price with everything switched on and the migration done for you.

A blunter version: Spott is the better choice for an agency that wants more of the recruitment workflow inside one tool. Kepler is the better choice for an agency that thinks its problem is not the workflow but the database underneath it.

Can you migrate from Spott to Kepler?

Yes, and the migration is done for you on every Kepler plan rather than handed over as a spreadsheet and a login. You send an export from Spott, plus any loose CVs, spreadsheets and shared-inbox archives that never made it into the CRM. Kepler handles extraction, mapping, de-duplication and validation, and rebuilds your custom fields as real fields rather than flattening them into notes.

Before the full load, you review fifty of your own records fully parsed — your people, not a demo set — and anything wrong is corrected and re-run. Activity, stages and history come across alongside the records themselves, so the desk keeps its memory. Kepler also exports the entire workspace on request at any time, with no exit fee, which is the same courtesy in the other direction.

The reverse move is also available: Spott publishes import tooling and onboarding documentation, and both products have a public API.