ATLAS is RoleMapper's AI agent, built into the platform as a slide-out chat panel or it's own page. It understands natural language — which means users don't need to know the exact title or reference number of what they're looking for. They can describe a role, a situation, or a question, and ATLAS will work out what to surface on their behalf.
Unlike a traditional keyword search, ATLAS can handle imprecise language, interpret context, and return results with an explanation of why they match — helping users make faster, more confident decisions without needing to escalate to a specialist.
The scenarios below cover the most common situations users find themselves in. For each one you'll see the goal behind the search, example prompts, and what ATLAS returns.
Finding a Document I Know Exists
GOAL
You need to retrieve a specific document you know is in the system — to assign it to a hire, confirm its status, or use it as a reference in a compensation decision. You're not browsing; you know what you want.
EXAMPLE PROMPTS
"Sales Manager Level 6 in Draft status"
"RM-0013452"
WHAT ATLAS RETURNS
Match found - ATLAS returns the profile directly — title, level, job family, and status visible at a glance. Filters like status or level scope the result further if included in the prompt.
Reference lookup - A direct reference number (e.g. RM-0013452) bypasses all inference and pulls the exact record - useful when working from a spreadsheet or system export.
I Know What The Role Does, But Not What We Call It
GOAL
You are working from external language - a candidate's CV title, a job board description, or market terminology - and need to find the internal equivalent. A traditional keyword search would fail here.
EXAMPLE PROMPTS
"someone who manages employee learning programmes"
"head of people ops"
WHAT ATLAS RETURNS
No exact match - ATLAS confirms it couldn't find a title match, then surfaces the closest equivalents from the catalogue - e.g. 'Information Security Officer' or 'L&D Manager' - with a brief note explaining the match.
Terminology bridge - This is the scenario where traditional search breaks down entirely. ATLAS maps the intent of the prompt to the nearest internal equivalent without requiring the user to already know the organisation's vocabulary.
Getting A Landscape View Of An Area
GOAL
You aren't sure yet which profile is right — or even whether one exists. You want to understand the full picture across a function or level range before making a decision.
EXAMPLE PROMPTS
"What Project Manager profiles do we have?"
"How many levels of Software Engineer exist in our architecture?"
WHAT ATLAS RETURNS
Grouped results - ATLAS returns a structured view of what exists — levels represented, job families, track spread - rather than a flat list the user has to interpret themselves.
Clarifying question - Where the breadth of results is significant, ATLAS asks a clarifying question to help the user narrow focus - for example, 'Are you looking for an IC or Management track profile?'
Checking If A Document Exists Before Creating One
GOAL
You are about to create a new profile and want to confirm nothing equivalent already exists. Getting a clear 'no match' answer is as important as finding a match — this is a pre-creation due-diligence step that protects the architecture.
EXAMPLE PROMPTS
"Do we have anything for a Data Governance Lead?"
"We're creating a new Procurement Manager role — does one already exist?"
WHAT ATLAS RETURNS
Match found - ATLAS confirms the profile exists and surfaces it — including whether it's approved or in draft. The user can use or copy it rather than creating a duplicate.
No match - ATLAS confirms clearly that nothing exists. It may also surface adjacent profiles in related families as a starting point, reducing the risk of near-duplication when building from scratch.
Comparing Profiles To Make A Decision
GOAL
You have identified that multiple profiles may be relevant and need to understand the differences between them - to place a person correctly, resolve a pay query, or support a compensation review.
EXAMPLE PROMPTS
"What's the difference between a Senior and Principal Engineer profile?"
"I need to compare our IC and Management track Product Manager profiles"
WHAT ATLAS RETURNS
Side by side - ATLAS returns the relevant profiles with key distinctions highlighted - level, track, scope of accountability - so the user can assess differences without manually cross-referencing records.
Next step prompt - Where a classification decision still needs to be made, ATLAS can direct the user toward the Classification Journey within the platform - keeping governance intact.
Benchmarking A New Hire's Grade And Pay Range
GOAL
There's a live offer or hire in progress. You need the approved profile and associated grade to make a defensible compensation decision — quickly, and with confidence the data is current.
EXAMPLE PROMPTS
"What level is our approved Procurement Manager profile?"
"We're making an offer for a Senior Data Analyst — what grade should we be using?"
WHAT ATLAS RETURNS
Approved profile - ATLAS returns the approved profile with level and grade information in context — ready to inform the compensation decision directly.
Status flag - If the profile exists but is in draft or unapproved status, ATLAS flags this clearly so the user doesn't rely on unverified data in a live offer situation.
Confirming That Something Doesn't Exist
GOAL
You need a definitive answer, not a guess. Confirmation that a role isn't in the system is itself useful information — it tells you to escalate, create a new profile, or go back to the requester with clarity.
EXAMPLE PROMPTS
"Veterinarian"
"Do we have a Quantum Computing Specialist profile?"
WHAT ATLAS RETURNS
Clear no match - ATLAS confirms that no match or close equivalent was found. It does not fabricate a plausible result or return a misleading near-match — a deliberate design decision.
Adjacent context - Where relevant, ATLAS surfaces the nearest job families or structural areas in the architecture so you have a starting point — whether that's creating a new profile or understanding where this role might sit.
Comments
0 comments
Article is closed for comments.