Skip to main content
Back to careers

Working at Atlan in the AI Era

What we expect. Why it matters. How it connects to how we perform.

Something fundamental has changed about what valuable work looks like. Output is no longer scarce. A competent skill can produce a first draft, a data analysis, a working prototype, an autonomous workflow. What remains scarce is the quality of judgment, the impact of the work, and the reach you create for yourself and the people around you.

We are not a SaaS company adding AI features. We are an AI-multiplied company: every role is expected to use AI to multiply what they do. This document makes our expectations explicit. They connect directly to how we assess performance. At the highest level, we need to be able to answer yes to these three questions:

Has AI expanded what I am capable of, not just what I produce, but the complexity and ambition of the problems I take on?

Has AI created leverage beyond myself: skills, systems, and knowledge that others build on?

Am I learning and rebuilding continuously, staying at the frontier of what is possible so I can keep elevating how I perform on the two questions above?

See our levels of AI nativity Go to the AI Native Bar ↓
Section 01

The baseline: what we expect from every Atlanian

Regardless of role, there is a set of expectations that are not optional.

Use AI in your daily work

You are using them to think, draft, analyze, automate, build, and improve — every day. If you are not, that is a gap, and it is visible.

Understand how AI tooling works well enough to assess the output

Usage without judgment is not the bar. You know when AI produces something good and when it produces something that is merely adequate. You push back. You refine. You do not ship the first output because it looks clean.

Be the human on the loop, not in it

Am I staying involved in this manually out of habit, or because my judgment is genuinely irreplaceable? AI handles the repetitive, high-volume, pattern-heavy work. You set direction, evaluate quality, handle exceptions, and make calls.

Continuously rebuild your own workflows

The tools change. Better approaches emerge. What was a strong workflow three months ago may not be the strongest one now. The expectation is not just to learn — it is to unlearn and rebuild, on a shortening cycle.

Share what you learn — and build on what others share

When you find something that works — a skill, a workflow, a new use case — you bring it back to the team. AI nativity compounds when it is collective. This runs both ways: share what you learn, and actively build on what others share.

Section 02

How expectations scale with your role

The baseline above applies everywhere. What scales is the depth of transformation we expect.

Roles like engineers, product managers, designers

You build what Atlan sells and ships. Your work is the platform. The AI-native expectation here is deep: zero hand-crafted code as the standard, AI embedded in every stage of the SDLC, judgment about what to build and what to delegate to agents. You are not just using AI to go faster — you are changing what "building" means.

The progression

  • Use AI tools fluently in your daily work
  • Understand the output well enough to evaluate, challenge, and improve it
  • Identify what in your domain should be augmented or automated — and say so
  • Contribute context, judgment, and requirements when others are building for your domain
  • At the high end: prototype, build, and drive adoption yourself
Section 03

Our AI Native Bar

Level 1: Below the bar

This is not a starting point aligned with being an Atlanian. It is a gap with a short-term clock on it.

Show
  • AI use is occasional and narrow: same tools, same prompts, same outputs for months.
  • Workflows look the same as they did a year ago. AI is additive at best, invisible at worst.
  • Output shows little evidence of interrogation: what gets produced is what the model returned first; "slop" is prevalent. When asked what they challenged, refined, or rejected, cannot answer specifically.
  • Has not actively retired or rebuilt a single workflow in the last three months. When better tools or approaches emerge in their domain, they observe rather than experiment and adopt.
  • Talks about AI in conversations but cannot point to decisions or work products it materially changed (abstract or performative, mostly hype and idea sharing, not application or depth).
  • Does not self-direct their own learning: not tracking what is changing in the field, not absorbing what colleagues are building, waiting to be told rather than seeking out what is new.
  • For managers: their team's operating norms have not visibly shifted, and they have not held their team to this bar.

Level 2: Augmented

The Atlan baseline. The floor, not the goal. If you are here and not moving, you are falling behind.

Show
  • Uses AI daily across core tasks: drafting, researching, analyzing, building, preparing. Every time, not occasionally.
  • Evaluates output critically: pushes back, refines, does not ship the first draft because it looks clean.
  • Has stepped back from pattern-heavy manual work: if AI can do the first pass, AI does the first pass. The question is no longer "should I use AI here," it is "why would I do this manually."
  • Work that remains manual does so for a real reason they can articulate: "This should still be manual because of X" rather than defaulting to "I'll just do it myself."
  • Begins to curate and structure their own domain knowledge (docs, notes, recurring context) in a form AI tools can actually use, not just in a form humans can read.
  • Practices regularly challenging constraints that limit workflow automation. Continues to experiment with new tools and approaches as the landscape shifts; does not wait to be told what to adopt.

Level 3: Integrated

The expectation for platform and customer roles. The floor for systems and tools roles.

Show
  • AI has changed how they work across their whole role, not just in pockets.
  • Has rebuilt how they approach at least one core part of their role end to end. Not just faster, structurally different.
  • Actively identifies what in their domain should be automated or redesigned, and says so, with specifics. Regularly refreshes this list.
  • Contributes context, judgment, and requirements when others are building tools for their domain. Not a passive consumer of what gets built.
  • Has shipped at least one AI-native workflow in the last two months that at least one other Atlanian depends on.
  • Understands the difference between raw information, knowledge that lives in people's heads, and machine-readable context; actively works to close that gap in their domain.
  • Shares workflows and discoveries with the team. Their AI fluency is raising the floor around them, not just improving their own output.
  • Scope and output are visibly larger: covering more ground, moving faster, producing higher quality than was possible before.

Level 4: Multiplying

The frontier. The expectation for systems and tools roles.

Show
  • Is running multiple workstreams in parallel at a scope that would have required a team before: AI handling execution, them providing judgment, orchestration, and quality. The output is visible and disproportionate.
  • Has rebuilt their entire approach multiple times as tools evolved. What they were doing six months ago is not what they are doing now, and they can name what changed and why.
  • Actively retires workflows that are no longer the strongest available approach, even when they still work fine.
  • Their AI fluency accelerates others: skills and workflows they build get adopted, and what they share raises the team's floor and breaks through ceilings.
  • Pursues outcomes that would have been considered out of scope before, and delivers them.
  • For systems and operations roles: the function's capacity grows without growing headcount. When something still runs manually, they can explain exactly why human judgment is irreplaceable at that step.
  • Has architected workflows where AI agents operate with meaningful autonomy, and has built the guardrails, checkpoints, and failure-mode awareness to make that safe and durable.
  • Converts tribal knowledge in their function into structured, reusable context that agents and teammates can both rely on. Knowledge that lives only in their head is a liability they are actively reducing.
  • Other Atlanians look to them to understand what AI-native work actually looks like in practice.
Section 04

What quality means now

The failure mode we are guarding against isn't someone who ignores AI. It's someone who is confident, polished, and mediocre — someone who produces clean-looking output and never pushes past good enough to great. Quality has two dimensions:

Judgment: did you bring something only a human could bring?

High-quality work reflects genuine thinking where you challenged the assumptions, pushed past the first draft, applied real domain knowledge, and produced something that AI alone could not have shaped to be excellent.

Ask yourself

  • Did I interrogate this output or accept it because it looked complete?
  • Did I bring real judgment, domain knowledge, or insight that changed the result?
  • Would a thoughtful person in this domain find this genuinely great — or just adequate?

Impact: did it actually move something to a material degree?

Good judgment applied to the wrong problem, or to a problem AI could have handled entirely, is not high performance. Quality also means working on the right things, using AI's leverage to do more, reach further, and drive bigger outcomes.

Ask yourself

  • Did this move something forward, or does it just exist?
  • Did I use AI's leverage to expand what I could accomplish, or just to do the same faster?
  • What changed — for the customer, the team, or the company — because of what I did?

Someone operating at the frontier has rebuilt their entire workflow as many times as needed, is working at a scope that would have required a team not long ago, and is actively pulling others forward — sharing what works, raising the floor for everyone around them.

Section 05

The mindset underneath all of this: unlearn and rebuild

There is a specific disposition that makes everything above possible: the willingness to let go of what has already worked.

AI capability has a shortening half-life. The tools available today are not the tools we will be using in six to twelve months. The workflows that are strong now may not be strong then. Someone who built a genuinely excellent AI-native workflow last year and has not touched it since is likely already behind.

The expectation is not just to learn. It is to unlearn and rebuild, on a cycle that keeps shortening. That requires intellectual honesty about when your current approach is no longer the best one.

Section 06

How this connects to performance

These expectations are not separate from how we assess performance at Atlan. They are part of it.

How you work with AI — the quality of your judgment, the impact of your work, the depth of your adoption, the degree to which you are continuously improving — is part of what it means to do your craft well at Atlan today. It shows up in craft mastery. It shows up in your outcomes. It shows up in whether you are making others better.

In Growth Conversations and Performance Reviews, your manager will be assessing not just whether you use AI tools, but whether AI is genuinely changing how you work. There is a meaningful difference between "AI is assisting me" and "I have deployed AI to radically multiply myself." The bar we are setting is the latter.

Explore open roles