It's the start of the quarter. You've been told your team needs OKRs, you understand the concept, and then you open a blank doc and freeze.
What do you write? How ambitious is ambitious enough? Is this even the right objective?
Good news – in a post-AI world, these questions can be quickly answered.
The blank page problem is solved
The hardest part of writing OKRs isn't understanding the format. It's knowing where to start. AI tools like ChatGPT or Claude are surprisingly good at getting you unstuck fast.
You don't need a perfect prompt. Just tell the AI what your team does, what you're trying to achieve this quarter, and ask it to suggest a few OKRs. It'll give you something to react to, which is 10x easier than writing from scratch.
A simple prompt like this works well:
I manage a 6-person product team at a B2B SaaS company. We're focused on reducing churn this quarter. Can you suggest 2-3 OKRs for my team, with 3 key results each?"
You'll get a draft in seconds. It won't be perfect (it'll probably be a bit generic) but it breaks the paralysis. From there, you edit, refine, and make it yours.
Anatomy of a good prompt for OKRs
The quality of what you get back depends on what you put in. A vague prompt gets vague OKRs. A specific prompt gets something actually useful.
A good OKR prompt has four ingredients.
1. Your team's context. Tell the AI who you are and what your team does. "I manage a product team" is fine, but "I manage a 5-person product team at a B2B SaaS company focused on mid-market customers" is better. The more specific, the more relevant the output.
2. Your timeframe. OKRs are quarterly by default, but say it explicitly. "For Q3" or "for the next 90 days" anchors the AI to goals that are actually achievable in that window, rather than multi-year ambitions dressed up as quarterly targets.
3. Your focus area. What's the one thing you most want to move this quarter? Reducing churn, growing revenue, improving onboarding, shipping a new feature — name it. If you have two or three priorities, list them, but don't go beyond that or you'll get a bloated list of OKRs that covers everything and focuses on nothing.
4. The format you want. Ask for a specific number of objectives and key results. Something like "2 objectives with 3 key results each" gives you a manageable starting point. If you don't specify, you might get 6 objectives and 18 key results, which is too much to react to sensibly.
Putting it together, a solid prompt looks like this:
I'm a team lead managing a 6-person customer success team at a B2B SaaS company. For Q3, our main priority is reducing churn among our mid-market accounts. Can you suggest 2 OKRs with 3 key results each? Key results should be measurable and achievable within 90 days."
That's it. No magic words, no special syntax. Just enough context for the AI to give you something worth reacting to.
One more tip: if the first output isn't quite right, don't start over. Tell the AI what's off. "The key results are too output-focused, can you make them more outcome-based?" or "The targets feel too conservative, can you push them a bit?" Iteration is fast and you'll get to something solid within a few rounds.
What to do with what the AI gives you
Your first set of OKRs will be wrong. They'll be too vague, or too ambitious, or focused on the wrong things. That's fine. That's expected.
The real value of using AI here isn't to get perfect OKRs on the first try. It's to give you a starting point you can pressure-test. Read the AI's suggestions and ask yourself:
- Does this objective actually reflect what we care about most?
- Can we realistically move these numbers in 90 days?
- Are we measuring outcomes, or just activities?
That critical thinking is still yours. AI just saves you the 30 minutes of staring at a blank screen.
The problem with copy-pasting from Claude or ChatGPT
If you use a general-purpose LLM, you still have to figure out what to do with the output.
You copy the OKRs from the chat window, paste them into a spreadsheet or a doc, share it with your team, and then... hope everyone remembers to check in on it. Most teams don't. The OKRs become a one-time exercise instead of a living plan.
That gap between "writing OKRs" and "actually running them" is where most teams fall apart.
A faster path: AI Mode in Tability
Tability has a built-in AI Mode that works the same way. You describe your team's goals, it generates OKR suggestions, but with one key difference: you can turn that output directly into a tracked plan.
No copy-pasting. No reformatting. No "I'll set this up properly later."
You go from a prompt to a live plan with owners, check-in reminders, and progress tracking in a few clicks. It removes the friction that usually kills OKR adoption before it even starts.
For team leads who are new to OKRs, that matters a lot. The fewer steps between "we agreed on goals" and "we're actively tracking them," the more likely your team is to actually follow through.
Which approach is right for you?
If you just want to explore what your OKRs could look like, or if you're still learning the format, start with Claude or ChatGPT. They're free, fast, and great for drafting.
If you're ready to actually run your OKRs with your team, use Tability's AI mode. You'll get the same quality of suggestions, but you won't lose momentum between writing and tracking (you can start a free trial at https://ww.tability.io).
Either way, stop waiting for the perfect goals. Write something, react to it, and improve it over time. Momentum beats accuracy every time.