Working with Andreas

If you’re working at Elicit, it’s because I think you are among the best people for your role. I want to do whatever will most help you succeed at it. You should treat me as a resource and think about how to make best use of me.

This document is a collection of policies, heuristics, and tactics that I think will help with that. They are instrumental, not an end in itself. I am happy to diverge from them if it helps you accomplish your goals at Elicit.

Unless I’ve heard otherwise from you, I’ll assume that you think the policies in this document are good for you and I will be surprised if we don’t implement them.

Table of contents

  1. How I work
    1. Do-acracy
    2. No task is beneath me
    3. Prototyping
    4. Using me well
    5. Known blindspots and tendencies
  2. Working together
    1. Your responsibilities
    2. How I can help
    3. How I should feel about your project
  3. Staying in sync
    1. Questions we should always be able to answer
      1. Big-picture questions
      2. Personal career questions
      3. Project questions
    2. Feedback
  4. Communication
    1. How I think and communicate
    2. Meetings
    3. Writing
    4. Email, Slack, Text
  5. Values
    1. Ambition
    2. Soul
    3. Goodness
    4. Truth
  6. Aesthetics
  7. Credits

How I work

Do-acracy

I prefer a do-acracy mindset. If you think something should be done and you can do it, just do it. You don’t need to ask permission for most things.

No task is beneath me

I’ll do whatever helps the team succeed, regardless of whose “job” it is. I expect the same from you.

Prototyping

I get energy from building prototypes. If you want to explore an idea together, building something is often more motivating for me than discussing it abstractly.

Using me well

  • I’m a strong generalist - I can help with engineering, design, ops, research, sales
  • I’m an underutilized teacher - if you want to understand something technical, ask me
  • Prompt me with good questions - I have lots of unspoken, unconventional beliefs that surface when asked the right questions
  • If you’re feeling pessimistic about Elicit, a 15-minute vision conversation with me might help

Known blindspots and tendencies

I’m working on these, but you should know about them:

  • I’m naturally agreeable. Despite my role, I find it hard to give direct negative feedback and start uncomfortable conversations. I’m working on this. If you want honest critical feedback, ask explicitly and I’ll try harder.
  • I can be a last-minute person. I naturally sprint toward deadlines rather than working steadily ahead of time. This doesn’t work well for things like fundraising or long-term planning. If you need something from me with lead time, build in buffer.
  • Ideas don’t always become action. I generate a lot of ideas but don’t always follow through on making them happen. If I suggest something and it matters to you, ask me to commit to concrete next steps - or feel free to let it drop.
  • I can be too prescriptive at the wrong time. I sometimes give detailed feedback on half-baked work when curiosity questions would be more helpful. If I’m jumping to solutions too early, tell me.

Working together

Unless we’ve agreed otherwise (or I’m not your manager), you are the owner of your project and I will try to help you with prioritization and resources, but expect a lot of autonomy from you.

Your responsibilities

You are the owner of your project. This means:

  1. I will defer to you on most project decisions, even if I disagree. I will communicate my reasons for disagreement
  2. You think about your project in the context of making Elicit succeed, and adapt it as needed
  3. You have some mechanism for tracking your goals, progress, and plans, and expose it to me
    1. For example, this could be a periodically updated spreadsheet that records what will happen when, what the visible results will be, and how long different parts will take
  4. You know your top 3 priorities at any time
  5. You know what it means to do a bad or mediocre job at the project so that you can check whether you’re in the zone of mediocrity
  6. You make tough choices about what to prioritize; you don’t just react to things coming your way
  7. If other people working on your project are slow to get work done, or do low-quality work, you fix it

How I can help

  1. In the beginning of our work relationship, I want to evaluate some work in great detail to align our expectations about what work should be like and to build trust
  2. I can help with prioritization. To do this well, it helps if I know:
    1. How much time are you spending on different activities now? Consider time tracking
    2. What are you currently prioritizing and deprioritizing?
  3. I can help with thinking about how your work fits into the big picture, and how to make it most useful for Elicit’s longer-term goals
    1. If in doubt, ask for more context, more often than you think you should
  4. I can help you coordinate with other projects at Elicit and elsewhere
  5. I am happy to consider any task that you think I’m in a better position to do than you
    1. But sometimes I won’t do tasks that you assign me
    2. Don’t assume that I know what is important; be explicit about importance
    3. If you gave me an important task and I haven’t done it, ping me about it
  6. I can brainstorm ideas with you, or help you think through plans and considerations
  7. I can make resources available for your project (hires, funding, connections) but you need to ask

How I should feel about your project

  • You’re doing whatever it takes to make your project succeed (within reason, and within the broader set of priorities of the organization)
  • There is very little room for feedback
  • You’re always a step ahead of me
  • I never have to worry about this again
  • I can easily summarize your progress to others
  • I have no doubt about whether the work is good or not, it’s obviously good
  • I want to give you more responsibility

Staying in sync

The most important thing about our working relationship is that we agree on what we are trying to do and how it’s going, on a personal and project level.

Questions we should always be able to answer

We should always be able to give roughly the same answers to the following questions. If you suspect that we have diverged, discussing these questions is our priority.

Big-picture questions

  • What is Elicit trying to do in the next 6 months, 1 year, 2 years?
  • How does your work fit into this plan?

Personal career questions

  • How satisfied are you with your work, 1-10? What are the major factors that raise or lower this score?
  • What are your personal goals at Elicit? Do you feel like you’re on a good path to achieving them?
  • What are your strengths, weaknesses, and overall performance?
  • What are you trying to improve at?
  • What do you like most and least about working at Elicit?

I’m also happy to share my personal answers to these questions - ask me.

Project questions

  • What are your goals on a time scale of weeks and months?
  • How far along are we?
  • How much progress have we made in the last two weeks?
  • What are your top priorities?
  • What are your key challenges? How can I help you?

Feedback

  • I will try to give feedback unprompted, but I sometimes forget. If so, you should prompt me
  • I am happy to give feedback on anything. If you want feedback on a piece of work, ask me. I usually won’t have feedback ready that I haven’t told you and instead need to explicitly reflect
  • Feedback from you to me is as important as feedback from me to you. I will make mistakes and need to know what I should improve at

When I think about feedback to you, I ask myself the following questions, only moving on to the next one if I understand the answer to the previous one:

  1. Communication: Is it clear what you are working on?
  2. Prioritization: Given that it’s clear, are those the right things to be working on?
  3. Progress: Given that it’s clear & they’re the right things, how much progress has there been?
  4. Implementation: Given that it’s clear & they’re the right things & progress is there, do I have feedback on the details of the work?

When feedback helps vs. hurts: I sometimes give detailed feedback too early - on half-baked work where curiosity questions would be more useful. If you’re still exploring and want space to think, tell me. I’ll try to ask “what are you trying to figure out?” before jumping to “here’s what I’d change.”

Communication

Communication will work best if you assume that I’m slow, forgetful, lazy, and impatient. I will try not to be those things.

How I think and communicate

  • I process information asynchronously. When you share something complex, I may pause to think before responding. Silence doesn’t mean disengagement - it means I’m processing.
  • Feel free to interrupt my thinking with more context or clarifications
  • I try to be intellectually honest - I’ll tell you my confidence level and admit when I don’t know something
  • I prefer discussions focused on content and ideas rather than navigating feelings. You don’t need to worry about how to phrase things.
  • If I compliment your work, I mean it. I try to only say useful, true things.
  • Be concise
  • Put the main messages first
  • Think about what I need to know and don’t need to know
    • I will know much less about your project than you
    • I like to understand the most important mechanics of what you are working on, not just the high-level gist. This means that you need to be an effective teacher
  • Tell me how you’d like to communicate - when to meet, use email, Slack, etc.

A note on repetition: I ask you to be concise, but I’ve learned that my own communication often needs more repetition, not less. If something is important - strategy, priorities, decisions - I should be repeating it until you’re sick of hearing it. If you’re unclear on our direction, that’s probably my fault for not repeating it enough. Ask me to clarify.

A note on power dynamics: As CEO, my suggestions carry more weight than I intend. What feels like brainstorming to me can feel like a directive to you. I’m often not aware of this dynamic in the moment. Please know: when I share an idea, it’s usually just an idea - not an instruction. You can push back, disagree, or ignore my suggestions. If you’re unsure whether something is a suggestion or a priority, ask me directly. I’m working on being more explicit about when I’m directing vs. brainstorming. Help me by asking clarifying questions.

Meetings

  • We have a running 1-1 doc. Most of the content in this doc should come from you
  • Prepare an agenda and share it with me at least 2h before each meeting
    • I will have better thoughts if I can think about your questions offline
    • It’s helpful to have a list of things we could talk about so that we can deliberately choose whether to discuss them or not
  • Our agendas should be decision-driven: “Here are the choices I’m trying to make. Here are the considerations I see. Here is my recommendation and rationale.”
    • Always provide defaults
  • Assume that I don’t remember what we discussed in our last meeting and that I need to be reminded of the context of your work. This isn’t because I don’t care, it’s because I have many different projects to keep track of.
  • If we’re spending a lot of our meeting time on trying to reconstruct what work you’ve been doing, or what candidates for next steps are, you didn’t prepare the meeting enough
  • I’ll often pause for a while in the middle of a conversation to think about what we’re discussing. Silence doesn’t mean that I want to move on to the next topic. Feel free to add more context while I’m thinking.
  • After the meeting, summarize the action items for you and me
  • If you want to meet outside of our regularly scheduled 1-1s, just add meetings to my calendar by making an event and inviting me. I’m basically always happy to chat about big picture things, ideas, hopes, despairs, etc.
  • I enjoy meetings that feel inspiring and energetic. What this can feel like:
    • Yes-and - we’re building off of each other
    • Excitement - you’re mirroring when I’m excited, and vice versa, we’re not dampening each other’s excitement
    • Optimism - we’re in a great place to do this, we can accomplish great things
    • Abundance mindset - so much possibility before us, so much low-hanging fruit
  • I strongly dislike: lack of agenda, meandering speeches that don’t move things forward, meetings that could have been async

Writing

  • Writing is a product, your readers are your users
  • All documents should have a <1 page summary. Most of the time I will only read the summary, and maybe zoom in on one or two sections
  • Structure writing hierarchically
  • Use lists, mostly ordered
  • Avoid jargon

Email, Slack, Text

  • Use Slack for tasks I can complete in <2 minutes, and for semi-urgent requests
  • If I send you a request, acknowledge it
  • For tasks that are not urgent and can’t be completed quickly, it’s best to communicate them to me via email or to batch them and tell me in our next 1-1
  • If something is urgent, text me

Values

Here are my core values at work. If I act in ways that are inconsistent with these values I want you to point it out to me. If you want to you can share your values with me and I’ll try to hold you accountable too.

Ambition

  1. Work hard
  2. Take ownership
  3. Be patient, long-term oriented
  4. Craftsmanship
  5. Challenge yourself and others
  6. Growth
  7. Against
    1. Risk aversion
    2. Easy wins
    3. Conformity
    4. Comfort

Soul

  1. Be human
    1. Authenticity
    2. Empathy
    3. Sincerity
    4. Integrity
  2. Be open and honest
  3. Share & elicit feedback
  4. Understand each other
  5. Don’t oversell, be humble
  6. Against
    1. Routine
    2. Hype without substance

Goodness

  1. Be impact-oriented
  2. Make the future go well
  3. Care about each other, community
  4. Abundance mindset

Truth

  1. First-principles thinking
  2. Rigor
  3. Curiosity
  4. Reflection
  5. Seeing with fresh eyes
  6. Be probabilistic, calibrated
  7. Go wherever reasoning and evidence take you
  8. Against
    1. Cached thoughts
    2. Black box thinking

Aesthetics

What I tend to appreciate - in work, ideas, and life:

  • Rigor + clarity: Clear thinking and rigorous execution. Transparency in reasoning (process over outcome). Making complex things accessible without hand-waving.
  • Competence and mastery: People being exceptionally good at things. Watching skill unfold. This applies to code, writing, thinking, and yes, movies like Moneyball.
  • Intellectual substance: Ideas at the center. I care about whether something rewards careful attention and respects the audience’s intelligence.
  • Systems thinking: Understanding how complex systems work and fail. Organizational design, feedback loops, seeing the mechanisms behind outcomes.
  • Meta/self-reference: Self-referential structures fascinate me - metacircular interpreters, programs that write programs, ideas about ideas.

Heuristics for whether I’ll like something (a product, a plan, a piece of writing):

  1. Does it reward careful attention?
  2. Is there rigor behind it?
  3. Does it show competence/mastery?
  4. Is it honest about complexity?
  5. Is there substance over style?

I avoid: vague hand-wavy thinking, bleak nihilism without resolution, slow meandering without purpose, style over substance.

Credits

  • Jungwon Byun
  • Jon Eng
  • Jay Desai’s user guide
  • GiveWell management docs
  • Maria Bridge
  • Elicit team
  • Birthday feedback from the Elicit team (2024)