Now that we are past Halloween, we can anticipate the most wonderful time of the year:
Annual planning season.
Admittedly, I may be an outlier here. I love this time and the opportunity for impact. A strong plan builds alignment and commitments. It sets your team up for a year’s worth of progress. It’s a rare time when executives focus on what needs to get done and the resources to achieve this. There is more flexibility than usual to revisit old assumptions while making new ones.
Many see this as another job on top of busy execution and an exercise to survive. Leaders plug in headcount growth of x% to match their budget and make up a BS roadmap with oddly specific dates, then teams create Gantt chart project plans. They are inventing a year’s worth of plans in a few days based on assumptions they neither understand nor document.
There are two likely (bad) outcomes of this process:
The makeshift plan becomes obsolete instantly, as there was never conviction to follow it.
The falsely precise plan is strictly followed, as sticking to commitments becomes more important than whether they were any good.
To call this “planning” is performative. Yes, there is a document created with that label. No, it is not helping you accomplish more. It could just as easily be created by AI. (Note: don’t ever have it built by AI.)
It doesn’t have to be this way.
A year is a long time and you should accomplish big things. Plan for it.
AI is changing a lot. Account for it. Here’s how.
Organize Around Durable Bets
In a world where anyone can build almost anything, it’s more important than ever that teams have a clear understanding of where the product is going. They also need to understand their role in the journey. An annual plan is the ideal time to reinforce these things, anchored in what is unlikely to change.
Some things to revisit:
Refresh your product vision: You should have a clear vision of the experience you are building towards for your customers. Even if it was “right” a year ago, update it and extend the time horizon.
Map teams to customer problems: Each team should have a charter that defines what part of the vision they are building and how progress will be measured. This gives them focus and limits conflicting changes that clutter the customer experience.
Plan roadmaps around themes: It’s not realistic to set 12 months’ worth of features and know that they will definitively solve a problem. Innovation doesn’t work that way, and assumptions change. Instead, plan strategic themes to address by quarter. Most features listed should be illustrative examples rather than commitments.
Features and requirements will change frequently. Your organization should provide direction and stability.
Staff to the Shifting Bottleneck
If you are getting productivity gains from AI, you probably have a new bottleneck: choosing what to build. Delivery is now significantly easier, which means feasibility isn’t your biggest problem. Turbocharging developers is probably squeezing discovery at a time when you need more of it.
Asking PM and design to just move faster is unrealistic, even with AI. Some things you can review:
Rebalance staffing by function: A product team that used to have 1 PM, 1 designer and 6-10 engineers may only need 2-5 engineers to deliver a similar backlog. Depending on your actual gains, that could mean smaller teams, more teams, or more PM and design capacity.
Collaborate continuously: AI enables people to be generalists. As roles blur, teams need to make ongoing decisions together to stay on the same page. Involve engineering in discovery and planning.
Raise the bar on assessment: When teams can ship faster, it becomes even more important to be selective about 1) what to build and 2) what to keep. Staff analytics accordingly and add rigor to your evaluations.
Staff and collaborate around current constraints.
Don’t Just Predict, Create
There is pressure to unlock big gains with AI. Some teams are reporting as much as 4-5x developer velocity when they embrace AI-driven development. However, improvements in outcomes are not keeping pace. If that’s true for your team, diagnose why rather than assuming next year will be different.
When you are forecasting what your team can achieve next year, make sure you have a plan to back that up.
Start from your current velocity: It’s tempting to make a big assumption about speed of development based on an external benchmark. Use current productivity as your baseline, then plan how you will improve it. If you’re just starting to adopt AI for development, be more cautious.
Plan for your product operating model: As Melissa Perri put it, “Whatever shape your operating model was in before AI arrived, AI is making more of it.” If you are a feature factory, AI will only make you a faster feature factory. Explicitly plan to improve prioritization and experimentation, in addition to how you deliver. Align around outcomes over outputs.
Budget capacity for change: Improvements can’t just be announced. They take focus, effort and training – all of which need management support and intentional team activities. Account for this in capacity planning to avoid overcommitting.
Make your plan a forecast, not a hope.
Hire for and Reward AI Fluency
AI is still new. There just aren’t many people who can claim to have years of experience. Your team should not be penalized for being new to using AI. However, that excuse will quickly become outdated.
As you are evaluating your team, making new hires, and reinforcing what’s important, building new skills should be high on your list. What this means specifically:
Find people who embrace learning: Being an expert in the old way of doing things has limited value when that way is becoming obsolete. A growth mindset brings the humility to recognize gaps and the effort to address them. Some refer to this as hiring for slope, not y-intercept. Look for it in hiring and reinforce it in performance management.
Provide support structures: Adoption will be faster if it’s shared and enabled. Whether it’s sponsoring training, creating forums for sharing, or doing brown bags, help people go through this process together.
Reinforce desired behaviors: Leadership needs to signal and encourage what is most important. An AI adoption mandate without follow-through gets quickly forgotten. Encourage experimentation, learning and sharing, even when attempts fail. As James Clear’s Cardinal Rule of Behavior Change puts it: “What gets rewarded, gets repeated.”
AI skills will keep changing. Create a culture of learning to keep up.
Plan for AI, But Not Just AI
Most of this advice isn’t new. AI is causing adjustments (e.g. staffing proportions) while amplifying other things that were already true (e.g. the importance of a product vision, a strong product operating model and hiring for slope).
The danger is that AI is also widening the gap between teams who do this well and teams who don’t. “Faster and focused” is competing with “throwing more spaghetti against the wall.” As always, be intentional and create the future.
As John Kotter says: “Leadership defines what the future should look like, aligns people with that vision, and inspires them to make it happen despite the obstacles.”
Before you finalize your plan, ask yourself: at the end of next year, how will your people, process and product be significantly better than today? This is our biggest job.
If you’d like a complimentary review of your 2027 plan, schedule time.
