AI reads the CV and scores it against the job description.
Why it exists Screens candidates in seconds instead of minutes.
Manual CV screening is the largest single time cost in recruiting and also the least consistent — the same CV gets judged differently at nine in the morning than at five in the afternoon, and reviewer number two applies a different bar than reviewer number one. AI analysis reads every CV against the same job description with the same standard, producing a score you can sort by. It is a triage instrument, not a decision: its job is to tell you which twenty of two hundred CVs deserve your actual attention. It is also the gateway feature for much of the product — profile enrichment, candidate facts and portfolio tabs all depend on it having run.
How to use it
Open the applicant's Overview tab.
Find the "AI Analysis Available" card.
Click "Analyze with CanviderAI".
Wait while "Analyzing CV..." runs — this takes a moment.
When "Analysis complete" appears, review the match verdict (excellent / good / fair / weak match).
Read the "Profile Match Analysis" score and summary.
Note: this costs 1 employer credit per analysis.
Good to know
Costs 1 employer credit per candidate. Use Bulk Candidate Analysis rather than clicking through profiles one at a time.
The score is only as good as the job description and skills it compares against — a thin posting produces meaningless scores.
Analysis also unlocks the Candidate Facts panel and the LinkedIn, ArtStation, Behance and Cara enrichment tabs.
Never reject on score alone. It is a ranking aid, and a human must make the decision.
Re-run analysis after materially rewriting a job description if you need consistent scoring across candidates.
AI-generated strengths and weaknesses for each candidate.
Why it exists Surfaces the decisive facts without reading the whole CV.
A single number tells you how a candidate ranks but not why, and "why" is what you actually need — to decide, to brief a hiring manager, or to plan an interview. Splitting the analysis into explicit strengths and gaps makes the reasoning inspectable, which matters for two reasons: you can sanity-check whether the AI weighted the right things, and you can defend the decision afterwards. The Shortcomings list is the more useful half in practice, since it tells you exactly what to probe if you decide to interview.
How to use it
Run AI analysis on the candidate.
Open the Overview tab.
Read the "Key Highlights" list for the candidate's strengths.
Read the "Shortcomings" list for gaps and risks.
Verify anything decisive against the actual resume.
Good to know
Shortcomings are gaps relative to this job description, not absolute judgements about the person.
These lists are the fastest raw material for a hiring-manager summary or an interview plan.
Verify any shortcoming you plan to reject on — AI misses skills that were phrased unusually in a CV.
A candidate with a mediocre score but highly relevant highlights is often worth a closer look than the ranking suggests.
Per-criterion pass/fail verdicts on your job criteria.
Why it exists Turns subjective screening into consistent, comparable checks.
This is the candidate-side view of the criteria you defined on the job, and it is where their value is realised. Instead of a blended score, you see discrete verdicts against explicit written standards — which is both faster to act on and far more defensible, since every candidate was assessed against the same published list. The confidence level attached to each verdict is the part most people overlook: a low-confidence pass usually means the evidence was ambiguous in the CV, and that is precisely the case to check by hand.
How to use it
Define criteria first via "Criteria Assertion AI" on the job details page.
Open the applicant's Overview tab.
Scroll to "Criteria Match Analysis".
Read each criterion's verdict and confidence level.
If it shows no results, re-run criteria matching from the job details page.
Good to know
An empty panel means either no criteria are defined for the job, or matching has not been re-run since they were added.
Low confidence is a flag to verify manually, not a soft fail — the CV was probably just ambiguous.
Once populated, the "Meets all criterias" filter on the People page turns these verdicts into a one-click shortlist.
Criteria assess evidence in the CV and profiles. A genuinely qualified candidate who did not mention something can still fail.
Suggested interview questions generated from the candidate's profile.
Why it exists Prepares targeted interviews probing real gaps and strengths.
Most interviews are under-prepared, because preparation happens in the ten minutes before the call and defaults to generic questions everyone has rehearsed answers for. Those interviews produce no information. Questions generated from this specific candidate's gaps against this specific role do the opposite — they go straight to the things you are actually uncertain about. This also improves fairness: candidates for the same role get probed on comparable dimensions rather than on whatever the interviewer happened to think of that morning.
How to use it
Run AI analysis on the candidate.
Open the applicant's Overview tab.
Scroll to the "AI Powered Questions" panel.
Read the suggested questions based on the profile and job requirements.
Copy the ones you want into your interview plan.
Good to know
Generated per candidate, so a shared question list will not match what these surface.
Best used alongside the Shortcomings list — the questions usually probe exactly those gaps.
Treat them as a starting set and add your own team-specific and role-specific questions.
Review before using: a question can be well-formed and still be irrelevant or inappropriate for your context.
AI proposes a better-fitting open job for the candidate.
Why it exists Rescues good candidates who applied to the wrong role.
Strong candidates routinely apply to the wrong opening — they pick the title they recognise, or the only role that was visible on the day. Rejecting them because the requisition does not match is a quiet, recurring waste, and one nobody notices because the alternative never surfaces. This checks the candidate against your other open roles and proposes a better fit, turning a rejection into a redirect. It works best when your open jobs have well-written descriptions, since those are what the comparison is made against.
How to use it
Open the applicant's Overview tab.
Find the "Suggested Position" panel.
Click "Suggest a better-fitting position".
Review the suggested job and the confidence level shown.
Click "View position" to inspect it, or "Refer to this position" to move the candidate.
Note: one-click referral needs the candidate's future-opportunities consent, and this feature requires employer tier 2 or higher.
Good to know
Requires employer tier 2 or higher.
One-click referral needs the candidate's future-opportunities consent; without it you must contact them first.
Suggestions are only as good as your other job descriptions — a vague posting will not be matched to well.
Worth running on strong candidates before rejecting them; it costs a click and occasionally saves a hire.
Always tell the candidate before moving them to another role.
Run AI analysis on every unanalysed candidate at once.
Why it exists Scores an entire backlog without clicking through each profile.
Analysis one candidate at a time is fine for a trickle and hopeless for a batch — after a bulk CV import or a good week on a popular role, you may have two hundred unscored applications, and opening each profile to click a button is an afternoon of nothing. Running the batch in one action is the difference between a scored pipeline you can sort and an unsorted pile you avoid. Do it right after any bulk import, so that the first time you look at the list it is already ranked.
How to use it
Open the People page.
Optionally filter to the job or group you want analysed.
Click "Analyze All Candidates".
Wait — progress is tracked and the page reports when the batch finishes.
Refresh the page if analysis is taking longer than expected.
Review the new Score column values.
Good to know
Each candidate costs 1 credit, so a 200-candidate batch consumes 200 credits. Filter first if you only need one role scored.
Only unanalysed candidates are processed, so re-running does not double-charge for people already scored.
Large batches take real time. Let it run rather than restarting it.
The natural sequence is Bulk Upload CVs → Bulk Candidate Analysis → sort by Score.
Side-by-side AI comparison of two to four candidates.
Why it exists Structures the final choice with a rubric and ranking.
The final decision between three strong finalists is where hiring is most vulnerable to bias and to whoever argues most confidently in the room. Individual scores do not settle it, because they were produced independently and are not calibrated against each other. A direct comparison with an explicit rubric forces the trade-offs into the open: this candidate is stronger technically, that one has the domain experience, here is what would change the ranking. The "What could improve ranking" output is genuinely useful beyond the decision — it tells you what to verify in a final interview or reference check.
How to use it
Click "Compare Candidates" on the People page, or open /ai-decision-helper/.
Choose the job in "Select position".
Tick 2–4 candidates from the loaded list.
Click "Compare".
Wait while "Analyzing candidates..." runs.
Read the TL;DR, Top pick, Weighted Score, Rubric Scorecard and Comparison Matrix.
Read "What could improve ranking" for each candidate.
Note: this costs 3 employer credits per comparison.
Good to know
Costs 3 employer credits per comparison, regardless of how many candidates you include.
Limited to 2–4 candidates, which is deliberate — comparisons across larger sets are not decision-useful.
Candidates should already be analysed; the comparison builds on their individual analyses.
The rubric scorecard is the part worth taking into a hiring meeting; the "Top pick" is an input to that discussion, not a verdict.
Results are saved and can be reopened later without paying again.
Reopen a previously generated AI comparison result.
Why it exists Revisits decisions without paying for a new comparison.
Hiring decisions get revisited constantly — an approver asks for the reasoning a week later, a preferred candidate declines and the runner-up comes back into play, or someone joins the panel late and needs to catch up. Regenerating the comparison each time would cost credits and, more subtly, could produce slightly different output, undermining the record of what was actually decided. Cached results keep the original analysis stable and free to reopen, which makes it a genuine decision record rather than a disposable one.
How to use it
Open the AI Decision Helper page.
Open the "Past comparisons" dropdown.
Select the comparison you want to reload.
The saved result renders, marked "Cached result".
Run a fresh comparison instead if the candidate set has changed.
Good to know
Reopening a cached comparison costs no credits.
The "Cached result" marker means it reflects the candidates as they were at the time, not as they are now.
If a candidate has since been analysed again or the shortlist changed, run a fresh comparison.
This is the cheapest way to answer "why did we pick them?" weeks after the fact.
AI writes candidate emails based on status and your context.
Why it exists Produces polished candidate communication in seconds.
Candidate emails are where good intentions die. Rejections in particular get postponed because writing something considerate takes effort, and postponed rejections become never — which is the single most common complaint candidates have about recruiters. Generating a draft removes the blank-page cost so the message actually goes out today. The 150-character context box is the important control: the model will not invent facts beyond what you give it, so telling it "strong portfolio, too junior for this role, encourage reapplying" produces something specific and honest rather than generic filler.
How to use it
Open the applicant's "Emails" tab, or the "Change State" modal.
Switch to the "AI Draft" mode.
Type optional context (up to 150 characters) — AI will not invent details beyond it.
Choose the length/tone setting (e.g. Medium / Balanced).
Click "Draft with AI".
Review the "AI Draft Preview".
Click "Apply to Email", edit if needed, then send.
Good to know
The context box is capped at 150 characters but is the highest-leverage input — a specific hint produces a specific email.
Always read the draft before sending. It goes to a real person under your name and cannot be recalled.
Available both from the Emails tab and inside the Change State modal, so the message can go out with the status change.
For messages you send repeatedly, build a saved template instead — it is faster and more consistent.
Automatic check that a description is complete enough to publish.
Why it exists Prevents thin or broken postings going live.
A published job is visible to candidates, syndicated to boards and sometimes indexed by search engines, so an accidentally empty or half-written description is a public embarrassment that is hard to fully retract. The check is a guard rail against the most common version of that mistake — clicking through the wizard faster than you were writing. It is worth being clear about its scope: it verifies that a description exists in sufficient quantity, not that it is any good. That judgement is still yours.
How to use it
Write your description on the "Job Description" step.
Watch for the notice "This description is not ready to publish yet".
Keep adding content — at least 50 characters is required to continue.
When the notice clears and "Save" enables, click "Next".
Good to know
The 50-character threshold is a floor, not a standard. Real postings need far more to attract candidates and to score CVs meaningfully.
The check measures completeness, not quality — it cannot tell you the description is boring or unclear.
If the notice will not clear, the content is likely below the minimum or the editor lost your text; re-check the editor body.
Because AI scoring compares CVs to this text, a minimal description directly weakens every later match score.
Card-based feed of AI-suggested bulk recruiting actions.
Why it exists Previews autonomous recruiting suggestions you can approve.
Today's ATS is reactive: it holds data and waits for you to notice what needs doing. Flow previews the opposite model, where the system watches your pipeline and proposes the actions — reject this stale group, chase these candidates, advance these strong matches — leaving you to approve or decline. The approval step is the design principle rather than a limitation: recruiting decisions affect real people's lives, so automation proposes and a human disposes. This is a proof-of-concept populated with sample suggestions, so treat it as a look at the direction rather than a working tool.
How to use it
Click your avatar in the navigation bar and choose "Canvider Flow".
Read each suggestion card: the proposed action, how many candidates it affects, and the reason.
Click the chevron to expand "Detailed Analysis" and the affected candidate list.
Click "Take Action" to run it, "Ignore for now" to postpone, or "Reject" to dismiss.
Click "Switch Back to ATS Mode" to return to the Dashboard.
Note: this is a proof-of-concept preview populated with sample suggestions.
Good to know
This is a preview with sample data — do not plan real workflows around it yet.
Always expand "Detailed Analysis" and check the affected candidate list before taking any action.
"Switch Back to ATS Mode" returns you to the normal Dashboard at any time.
Worth a look if you want to understand where the product is heading; skip it if you are here to get today's work done.