a

l

y

x

about

get in touch

alyx

a

l

y

x

about

get in touch

Generative AI for OKRs

SUMMARY

Writing is hard. Writing OKR's is particularly hard - especially when you have to consider your companies strategy, your teams core focus, and what you can humanly achieve in a time bound period. Gen AI now makes that a lot easier.

SUMMARY

Writing is hard. Writing OKR's is particularly hard - especially when you have to consider your companies strategy, your teams core focus, and what you can humanly achieve in a time bound period. Gen AI now makes that a lot easier.

What are we really doing here?


We've been studying & mapping the OKR creation process for some time - it's where teams often get stuck & the body of work an enterprise executes can become dramatically detached from its strategy. There was a clear opportunity for us to help users write impactful OKRs, and for company leaders to sleep better at night knowing their strategic goals flow into the rest of the organizations OKRs.


To do this - we had to get honest about the state of AI. This started with some questioning and analysis.


  1. What is AI presently good at?

    1. Summarizing mass amounts of data & finding themes/patterns within

    2. Generating answers from vague and sometimes unstructured thoughts


  2. Where does AI presently struggle?

    1. Accuracy; its confident but not always accurate

    2. It needs context and framing to give an answer in the right framework


  3. Where do humans get stuck in the strategy / OKR cycle?

    1. Finding themes / translating leadership priorities into outcomes they can can execute against

    2. Simply writing good OKRs (if you’ve been in an OKR setting session, you know!)

    3. Turning an outcome into tangible chunks of work to get started on

    4. Ensuring that an organization is aligned on strategy. No leader wants to spend money on work that doesn't drive the company forward.


These questions & conversations revealed a lot to us; most importantly, the need to help our users get over the writing block that many of them face, and to make sure that their OKRs are in line with the company strategy. We started to understand how thin the line is that we’re dancing on; help a user get 90% there, but make sure that their input is still required when interacting with AI. We want to boost the human brain, not replace it.


Important to note, this also strikes a happy balance with our customers & business process owners overseeing OKRs at the company. Leaders can trust that their teams are writing better OKRs in the context of their strategy, but they also know that AI isn’t just generating a slew of OKRs and calling it good.


User story map of OKR ideation process - and a few notes on where AI can play.

Our first dip into AI

We settled on enabling AI in our collaboration canvas to start - adding the ability for teammates to ask AI to help co-author OKRs. We went with 'atomic' interactions for speed and simplicity; clicking on an element enables you to work with AI to perform a task. For example, by clicking on an Objective, you have the opportunity to work with AI to draft or improve it.

Co-author is contextual to what you’re working on! Meaning you can use it for big ideas, or to fine tune small details. You’ll be able to find it in the toolbar of shells and various elements of the OKR shell.

Measuring Success

Gen AI has been in the product for 2 weeks - released to our beta customers just ahead of their OKR cycle for Q3.


What have we learned so far? That our prompt engineering is dead on. Customers are dazzled by getting over the writers block, and find that the responses returned are helpful; even if they're wrong. We've had teams share that it's a great place to spark conversation and removes any political barriers from the conversation.


We're tracking actions within experience; everything from draft, try again, and discard.