Write an Effective JD for AI Sourcing
Tips for describing roles to Mira's AI agent. Include implicit criteria like "startup DNA" or "built from scratch" for better candidate matching.
Mira's Agent understands natural language, you don't need to write a formal job description. But the way you describe a role significantly impacts the quality of your Shortlist.
Good vs. weak descriptions
| Weak | Good | Why it matters |
|---|---|---|
| "Find me a developer" | "Senior backend engineer, 5+ years Python, has built microservices at scale, preferably from fintech" | More context gives MRE more signals to match against |
| "Marketing person in NYC" | "B2B SaaS marketing manager in NYC, has done product launches, comfortable with data-driven campaigns" | Implicit criteria ("comfortable with data") triggers reasoning-based matching |
| "We need someone good" | "We need someone who's taken a product from 0 to 1, ideally at a Series A-B startup" | Mira interprets "0 to 1" as early-stage building experience |
What Mira Reasoning Embedding understands
Mira Reasoning Embedding goes beyond keyword search. You can use:
Arithmetic criteria
"At least 5 years of experience in backend development"
MRE calculates experience duration from the candidate's career timeline, not just from a profile tag.
Semantic criteria
"Someone with 0-to-1 experience" or "has built a team from scratch"
MRE interprets the implied meaning, early-stage company building, ground-up team creation.
Commonsense criteria
"Public company background" or "high-growth startup DNA"
MRE infers organizational context from the candidate's work history (company size, funding stage, growth trajectory).
Multi-source criteria
"Has sales experience at a public company and also has startup experience"
MRE performs multi-hop reasoning across different parts of a candidate's career.
Tips for better results
-
Be specific about skills, not just titles. "Full-stack" means different scopes to different teams. Say "React + Node.js + PostgreSQL" if that's what you need.
-
Include implicit criteria. The criteria you'd tell a recruiter verbally : "someone scrappy", "can operate independently", "has managed remote teams" : Mira understands these too.
-
Mention the company type. "From a YC startup" or "big tech background" dramatically changes who the Agent surfaces.
-
Specify location preferences clearly. "Berlin preferred, remote OK" is clearer than "flexible location."
-
Don't over-filter. Start broad, then refine. You can always tell the Agent "narrow it down to people with more startup experience" after seeing the first Shortlist.
What if the results aren't right?
After reviewing a Shortlist, you can adjust in the same conversation:
- "Show me more junior candidates."
- "Include people from Amsterdam too."
- "Focus on candidates from e-commerce companies."
- "I want people who are currently looking for a new role."
The Agent remembers the full conversation context and adjusts the next search accordingly.