An Overview of Ray: Sourcing Agent

The sourcing agent is currently in Limited Availability. Reach out to your system admin to get access.

The sourcing agent represents a shift from traditional search to Generative Powered Sourcing. Instead of manually building Boolean strings or configuring complex filters, Ray uses AI and natural language to automate the most time-consuming parts of candidate discovery — so your team can focus on high-impact candidate engagement.

Ray is an AI-powered sourcing assistant built into Beamery. It leverages Large Language Models (LLMs) and Beamery's proprietary Talent Match AI to understand the intent behind your hiring needs — not just keywords. Give it a brief description or a full job description, and it will build an intelligent, targeted search strategy for you.

Where to Find Ray

Ray is available on your Beamery homepage. Once you log in, you'll see the chat interface on your homepage — this is your starting point for every search. 

Watch a demo here:

Key navigation:

• New Search: Press the New button each time you're searching for a different role. This ensures Ray starts fresh with the right context.
• History: Every search you run is saved. Click the History icon to revisit any previous search and pick up where you left off.

How to Start a Search

Click Start New Prospect Search to get going. From here you have two options:

Keep it simple: Type a short request like: I’m looking for [role] in [location]. e.g. "Find me a sales specialist in New York."

Give it more context: The more information you provide, the more tailored Ray's search will be. Paste in a full job description, intake notes, or any other details about the role. Ray uses all of that context to tailor the search – this is how you can influence the search strategies generated.

Choosing a Role from Beamery's Taxonomy

After you submit your request, Ray will identify which roles in Beamery's taxonomy are most relevant. You'll see a short list of suggested roles — you can read the description of each one if the title alone isn't clear.

Pick the one that best matches what you're looking for. Don't overthink this step; it's not overly precise. Ray uses this selection as an anchor to identify the right family of job titles, and it will expand to related titles in the next step.

Understanding Search Variants

This is where Ray really saves you time. Instead of creating multiple searches yourself, Ray automatically generates two search variants, each with a deliberately different strategy. The two variants are designed to give you both a focused and a broad option so you can choose the right approach for your talent pool.

It’s a lot easier to review and adjust these pre-built searches than it is to build a search from scratch. Always review the filters that have been automatically generated – Ray does the best that it can based on the input given – but it’s your job to double check. 

Precise

The Precise variant is designed to return highly targeted results. Its philosophy is: direct synonym roles + core skills + ideal companies.

It includes:

• Role title variations (direct synonyms only): A focused set of closely matching job titles — direct synonyms and close seniority variations within the same discipline as the target role. Ray matches these against Beamery's taxonomy, so the titles are standardised and searchable. Ray searches across a contact's entire work history — not just their current role — so it catches people who have held relevant titles at any point.
• Skills (intentionally minimal): Ray deliberately selects only the most critical 0–5 skills from your job description — not all of them. The goal is to pick the skills that best differentiate the role without over-restricting results. For example, for a Frontend Engineer, Ray might select only "React.js" rather than listing React, TypeScript, JavaScript, CSS, HTML, and everything else. For some roles where the title alone is distinctive enough (like a Payroll Tax Analyst or a News Producer), Ray may apply no skills at all. These can still restrict your result count depending on data quality in your CRM, so if you're seeing too few results, try removing some or all of the skills.
• Location and radius: Ray identifies the location you specified, centres the search there, and sets a commutable radius. It uses its knowledge of the area to choose an appropriate distance — for example, it might set 40 miles for a city like New York, but adjust differently for a city with different commuting patterns. Location also influence the next step – Target companies.
• Target companies (tiered): Ray builds a list of companies where potential candidates might work, based on the locations you specified in the initial prompt. These lists are drawn from two tiers:
  • Direct competitors — the closest competitors in the same niche producing similar products or services.
  • Adjacent companies — same industry or sector but not direct competitors (supply chain partners, equipment manufacturers, related technologies).
  • Ray automatically excludes your own company and its subsidiaries, and includes common name variations (e.g., "Google", "Google Inc", "Google LLC") to improve matching.

When to use Precise: When your CRM has a large talent pool and you need to narrow down to the most relevant candidates quickly.

Roles First

The Roles First variant deliberately casts a wider net. Its philosophy is: direct and adjacent role titles, no skills, no companies.

It includes:

• An expanded list of role titles (direct + adjacent): In addition to the direct synonyms from Precise, Roles First adds adjacent role variations — related but different disciplines, transferable skill sets, common lateral moves or career pivots, and industry-specific terminology variations. For example, if you're looking for a Frontend Engineer, Roles First might also include Full Stack Engineer, UI Engineer, and similar adjacent titles.
• A wider location radius: Ray automatically increases the search radius by approximately 25% compared to Precise, giving you a broader geographic catchment.
• No skills filters: Roles First deliberately removes all skill filters to maximize the number of results.
• No company filters: Roles First deliberately removes all company filters to avoid restricting the pool.

When to use Roles First: When you're struggling to find enough contacts in your CRM, when you want to explore a broader set of candidates before narrowing down, or when you're open to candidates from adjacent roles who could be a strong fit.

Review the role titles in the Roles First variant and remove any that feel too broad. For example, if you're hiring a Sales Specialist, a title like "Investment Sales Broker" might be too wide — deselect it and update the search to refine your results.

How Ray Builds Its Search — Behind the Scenes

Understanding what Ray does under the hood can help you get more from it and know when to intervene.

Role Matching

When you describe a role, Ray generates two pools of job titles:

1. Primary roles (used in Precise): Direct synonyms and close seniority variations of your target role, all within the same discipline. These are reconciled against Beamery's taxonomy to ensure they're searchable.
2. Secondary roles (added in Roles First): Adjacent roles from related disciplines, lateral moves, career pivots, and industry-specific variations. These are also reconciled against the taxonomy.

This is why you might see 3–10 titles in Precise but many more in Roles First — this is by design.

Skills Selection

Ray doesn't just use every skill from your job description as filters. It analyses the role's seniority level and intentionally selects only the most differentiating skills — typically 0 to 3, and at most 5. The reasoning is that over-filtering on skills dramatically reduces results, especially when CRM data quality varies. For senior roles, Ray may prioritize strategic skills (like "System Design") over foundational ones (like "JavaScript"). For some roles, the title alone is distinctive enough that no skills are needed at all.

Skills are only applied to the Precise variant. Roles First always has an empty skills filter.

Company Generation

For the Precise variant, Ray generates a list of target companies organised into tiers based on any locations you have previously supplied:

• Direct competitors: Companies in the same niche producing similar products or services.
• Adjacent companies: Same industry or sector — supply chain partners, equipment manufacturers, related technologies.
• Cross-industry talent pools (considered but selective): Companies in different industries with overlapping job composition — for example, Data Scientists from finance for a tech role. Ray only includes these when the role's skills genuinely translate across industries, and skips them for highly specialized roles where domain knowledge matters.

Ray excludes your own company and includes common name variations and misspellings to improve matching. The Roles First variant deliberately has no company filter.

Your Edits Are Preserved

When you modify filters — either directly in the UI or via chat — Ray preserves your changes. If you ask Ray to refine a search, it builds on your current filter state rather than starting from scratch. This means you can safely iterate: adjust a few role titles, ask Ray to add companies, remove a skill, and each change is additive rather than destructive.

Interacting with Ray via Chat

One of Ray's most powerful features is that you can modify your search using the chat — just like you would with a tool like ChatGPT. You're not locked into the filters Ray initially generates. The workflow is flexible:

• Start in chat: Describe what you need and let Ray build the initial search.
• Adjust in the filters: Manually tweak the filters the way you're used to.
• Go back to chat: Ask Ray to make further changes using natural language.

You can keep going back and forth between the chat and the filters as many times as you need.

Examples of What You Can Ask

• "Can you add an extensive list of target companies please?"
• "Add these 15 skills: [list]"
• "Add these 52 roles from my Boolean"
• "Convert this Boolean into the filters"
• "Make the search more precise"
• "Remove the location filter"
• “Update all the filters in precise so they are focusing on talent with medical background'
• “Tell me about XY corp do they work in the medical sector”
• “Add cities in UK which will have mecanical engineering talent within a 4 hour drive of Bristol”
• “Find me talent in the top 20 local news stations”
• “Make all of the location radiuses wider by 20%”
• “Add an extensive list of role titles which would be relevant for this role in a medical tech sector”

Ray will update the search filters in real time based on your instructions.

Use the sourcing agent like you would any other AI chat interface, just ask.

Tips for Getting Better Results

• Paste in the full job description when starting a search. The more context Ray has, the better it can tailor role titles, skills, and company filters.
• Clear skills filters if your result count is too low. Skills depend on data quality in your CRM, and they can be the biggest limiter on results. Ray already keeps skills minimal by design, but sometimes even 1–2 skills can be too restrictive.
• Clear company filters if you're still seeing too few results after removing skills.
• Use Roles First when your talent pool is small or you want to explore more broadly before narrowing. Remember it already has a wider location radius, no skills, and no companies — so it's designed to maximise results.
• Use Precise when you have a large CRM and want to get straight to the best-fit candidates.
• Iterate using chat. Don't treat the first set of filters as final — keep refining by talking to Ray. Your edits are preserved between iterations.
• Check the role titles in Roles First carefully. Because it includes adjacent roles, some may be too broad for your needs. Deselect any that aren't relevant and update the search.

Taking Action on Results

Once you're happy with a search variant, review the candidates Ray surfaces. You can:

• Review matches directly within Ray's results view.
• Open your results in the People Grid for additional actions and more options for engaging with candidates.

Key Features at a Glance

• Automated Search Strategies: Ray automatically builds both a precise and an adjacent-role search for every role, removing the need for manual search configuration.
• Natural Language Interaction: Describe what you're looking for in plain language — no need for Boolean or complex filters. Ray handles the translation.
• Contextual Intelligence: Ray understands your company, your competitors, your industry, and commuting patterns for different locations. It uses all of this context to build smarter searches.
• Intelligent Skills Selection: Ray intentionally keeps skill filters minimal to avoid over-restricting results, selecting only the most differentiating skills for each role.
• Tiered Company Targeting: Ray builds company lists across direct competitors and adjacent companies, including common name variations for better matching.
• Edit Preservation: Your filter changes — whether made in the UI or via chat — are preserved when you iterate, so you never lose your refinements.
• Human in the Loop: Ray is designed to support your team, not replace it. No automated decisions are ever taken — Ray surfaces candidates and search strategies for you to review and act on.

Business Impact

By automating the candidate discovery phase, Ray reduces the time your team spends on manual sourcing. This frees you up to focus on higher-value activities such as:

• Candidate engagement and outreach
• Hiring manager calibration
• Building and nurturing talent pipelines

Additional Resources

Understanding Artificial Intelligence in Beamery
Beamery AI FAQ
Beamery AI Explained