Real candidates · AI-assessed · Verify in one click

Match your JD to vetted India developers.

Paste a job description. Get 5 real candidates from our pool of 1,200+ vetted developers — with grounded fit assessments, strengths, honest gaps, and live prices. About 10 seconds.

Try a sample
AI surfaces real candidates from our vetted pool. Each match is grounded in profile data. 5 JDs per IP per day.
The matcher

Three steps. No magic.

Careful prompting paired with a real, vetted catalog. No invented experience, no hallucinated names.

  1. 01

    We read your JD

    An AI pass extracts structured intent: role, required tech, nice-to-haves, experience band, industry exposure, must-haves, timezone.

  2. 02

    We narrow the pool

    Must-haves run as hard filters against 1,200+ profiles. Typically 30 clear; we then AI-score the strongest 8.

  3. 03

    We assess and stream

    Each candidate's profile JSON is read alongside your JD. Output: fit score, why-they-match, strengths grounded in the profile, honest gaps.

Different from asking ChatGPT.

Every match links to a live profile. Every claim cites profile data. Every card lists honest gaps.

Real candidates, not invented names

Every match links to a live profile page with resume, projects, and references. Audit our work in one click.

AI is forbidden to invent experience

The per-candidate assessment is grounded in the profile JSON. Strengths must cite specific fields. The model can't promote 'used React once' into 'React expert.'

Why-this-match framing for every candidate

Each card synthesises specific differentiators — rare skill combinations, ramp-time advantages, domain depth — grounded in profile facts. If the AI can't find genuine standout points, we lower the score and flag the match as medium-confidence so you know to dig deeper.

Prices match the profile page exactly

No hedge pricing, no 'starts at' disclaimers. The monthly USD figure is the actual contract rate for that specific person.

Common questions

The questions buyers actually ask on sales calls, answered honestly.

Can the AI invent experience that a candidate doesn't have?+
No. The per-candidate assessment is given the actual profile JSON and required to cite specific fields. Every 'strengths' bullet and every 'Why <Name>' differentiator must reference something in the profile. If the model can't find genuine standout points beyond "they know the tech", we lower the fit score and mark confidence as medium — that's the calibration signal that this is a workable match rather than a standout one.
How accurate are the matches?+
We expect ~80-90% of well-written JDs to yield at least 3 interview-worthy candidates in the top 5. The other 2 are usually weak-but-not-wrong (close stack, wrong experience level). The score range itself signals confidence: 85+ is strong, 70-84 is good, below 70 means verify carefully on the full profile. The 'View full profile' link is your safety net — full resume, projects, references — for any specifics the card summary couldn't capture.
Why am I only seeing 5 candidates? You said there are 1,200+.+
We pre-filter against your JD's must-haves (tech, experience band, industry) to narrow ~1,200 profiles down to ~30 that could plausibly match. Then we run AI assessment on the top 8 of those and return the 5 highest-fit. Showing more would mean reading through worse matches — better to surface 5 strong ones.
What if none of the AI matches are right?+
If under 3 candidates score above 70, the tool says so explicitly and offers to source to spec via the contact form. We'd rather be honest than waste your time with bad matches.
Is my job description stored or shared anywhere?+
The JD is sent to OpenAI for the analysis and cached on our server for 24 hours so re-submits don't cost extra. It's not stored permanently, indexed, or shared with developers. Treat it the way you'd treat any document you paste into ChatGPT.
How long does this take?+
Usually 8-15 seconds end to end. The first candidate card appears within 2-3 seconds. We stream results as each candidate's assessment completes so you don't sit watching a spinner.
Why is there a daily limit?+
5 JDs per IP per day keeps OpenAI costs sustainable and blocks scraping. If you genuinely need to match more in a day, the contact form route is faster anyway — a human on our side does it in 10 minutes per JD.
Can I hire on contract or do I need to commit?+
Witarist is contract-only. Monthly contracts by default, with hourly and project-based options. No fixed commitment — extend month-to-month, scale up, or end with 14 days' notice.

Prefer to talk to a human?

Send us your JD directly. A senior on our side reads it personally and replies with 3 hand-picked candidates within 4 business hours.

Send us your JD →Browse the catalog