Not every signal deserves your team’s attention. Avina’s Qualification Agent runs behind the scenes on every signal, evaluating accounts and contacts against your Ideal Customer Profile to determine who’s worth pursuing and who isn’t.
The result: your team only sees high-quality, pre-qualified leads in their feed, and automations only fire for accounts that actually match your target market.
What the Agent Does
Every time a signal fires, the Qualification Agent evaluates it across two dimensions:
- Account fit: does this company match your ICP?
- Signal relevance: is this the right person, triggering the right signal, at the right time?
Signals that pass both checks are surfaced to your team. Signals that don’t are suppressed or deprioritized, saving your team from noise and protecting your credits.
Account Fit (ICP Scoring)
The agent scores every account against your configured ICP and assigns an ICP Fit letter grade (A–F) plus a numeric Account Fit Score (0–100).
Signals from F-graded accounts are suppressed from notifications and outreach; signals from A/B accounts are prioritized in your Feed.
For the full breakdown of grades, how the score is computed, and how to configure your ICP, see ICP & Fit Scoring.
AI Signal Qualification
For signal types that support it, the agent goes beyond firmographic scoring and runs an AI-powered evaluation of each signal before it reaches your team.
This step uses an LLM with web search capabilities to assess whether a signal meets your specific requirements. You define these requirements in freeform text when configuring a signal (the “qualification” field), and the agent evaluates each potential signal against them.
How It Works
- Signal fires: a potential lead is detected (e.g., a new hire, a job posting, a social mention).
- Context is gathered: the agent assembles account info, contacts, engagement history, deal context, and your product positioning.
- AI evaluates: the agent assesses the signal against your qualification criteria using web research and contextual reasoning.
- Decision: the signal is either qualified (proceeds to summarization and delivery) or rejected (skipped silently).
Example
If you create a New Hire signal with the qualification prompt:
“Only surface hires at companies that have raised Series B or later funding and have over 100 employees”
The agent will research each candidate company, check funding history and headcount, and only deliver signals that meet those criteria.
Use the qualification field for criteria that can’t be captured by standard filters. If a filter already exists for something (like industry or location), use that instead for faster, more precise results.
Tuning the Agent
If AI-qualified signals aren’t precise enough:
- Add more specific criteria to your qualification text (companies, sizes, signals you care about).
- Be concrete about exclusions — tell the agent what not to surface as much as what to surface.
- Test with Preview before activating — the preview pane shows recent matches so you can verify the prompt is catching the right intent.
If too many or too few signals are coming through at all, the issue is usually your ICP, not the qualification prompt. See Tuning your scoring in ICP & Fit Scoring.
Best Practices
- Set your ICP first. The agent works best when your ICP is defined before you activate signals. See ICP & Fit Scoring.
- Gate outreach behind ICP Fit. Don’t blast sequences at every signal. Use ICP Fit grades as automation filters.
- Use qualification prompts for nuance. Standard filters handle the basics; qualification prompts handle the judgment calls web search can resolve.
- Combine fit with signal strength. The best outreach targets are accounts with both high ICP Fit grades and strong recent signal activity.
