Q3 OKR Draft
The problem: we are operating far below our potential. We have left several million dollars of economic value on the table. This re-org is the moment to fix that, and we have to take it: redefine and restructure around AI.
The good news is that we already have the part you can’t buy or build overnight. This is an incredible team, with the thing that matters most: good judgment, clear thinking, and a real vision for what’s possible for our product and our customers.
What’s missing is leverage. We have to hook that judgment up to the rocket of best-in-class AI practices, the thing that lets us move faster than we ever did before this re-org. Same team, same vision, multiplied.
This plan is how we become an AI-native company this quarter. For the full case, see the overview.
Binary goals
- 100% of the team is trained on how to use Claude Code to vibe code tools (in order to just get familiarity with Claude Code to enable the next step, which is what I care about)
- 100% of the company is trained on how to use Claude Code as an agentic workflow toolkit for their day-to-day work
- 100% of the company has built a workflow that enables their core work function
- 100% of the company has built a workflow that enables a secondary job function
Behavioral change
- 100% of the company has my sync tool running on their computer that pushes session data to a reporting platform (we can’t manage what we can’t measure, so we need this data)
- Whatever the baseline was at, we increase AI messages by 3x and increase the number of sessions that use a workflow resulting in markdown artifacts and intentional compaction by 3x
- Next quarter we have a very defined understanding of where individuals are falling short of where they should be, and a targeted training plan to help each one
Compounding outcomes
Every team and individual’s work shifts away from generating an output to generating systems that generate an output.
e.g. it is not an engineer’s job to write code, it is their job to create systems that generate clean code
- We work with each department head to set an output goal that would be impossible to attain without using AI
- For marketing and eng, I believe strongly that we need to, at a minimum, 3x the output per individual
- On eng, I’m confident here because I already spent a ton of time building a system for that purpose
- I’m not sure what it is for each other department, but it should be something along those lines
Dependencies
In order to achieve this I need to build the tooling:
- A CLI tool that can supply everyone’s sessions with the context they need (transcripts, analyses, other departments’ documents, etc.) to make everyone’s sessions powerful and up-to-date
- The session syncing tool that will be on everyone’s computers and syncs their Claude Code session data for analysis and measurement
- The context cortex that aggregates each department’s primary artifacts
- Trainings to:
- Get everyone familiar with Claude Code
- Teach how to use it as a toolkit for building your own agentic workflows
Stakeholders
- Andrew and I own this as an OKR or set of OKRs (using our combined technical, vision, mobilization, and change management skills), but with a very tight partnership with Chandler, who has driven so much of the context cortex and operational vision for enabling AI at Marq
- Every department head is a contributor to the outcome we are driving for (the impossible goal)
- Owen backs it up with a hard mandate and gives me a bark with teeth so I don’t waste my time building everything and get stuck