A few weeks ago I was speaking with a product manager of a well-known B2B2C company based in SF. We were talking about project and roadmap timelines and she let me know that often it takes them a year to go from feature request to feature delivered.
Withholding my gasp, I asked why.
She let me know that there are too many decisions to weigh, priorities to consider, work to do, and just too. much. stuff. 👏🏽 (and little to no processes to move through the stuff). In an attempt to be helpful, I told her about an acquisition where we needed to deliver technical field readiness faster than the historical timeline of 12 to 16 months. After I worked with the project teams on the framework for delivery they were able to realize this value in 4 months. In a whisper I heard her say, we should probably build some frameworks
That whisper told me two things: she wasn't proud of the one-year timeline, and she already knew what to do. What stuck with me from that conversation is that the bottleneck isn't the work, it's the friction between the people doing it. Too many handoffs, too many decisions waiting on the wrong person, too little clarity at the working level. So I went looking at the company most people point to when they talk about fast delivery: Anthropic. Luckily for me, many people are awe-struck by the speed of their delivery and in a podcast episode, this was the exact question asked of Cat Wu, Head of Product, Claude Code at Anthropic. She broke down their release flow and how they are able to move so crazy fast (from idea to delivery in a handful of days).
The 8 systems behind Anthropic's delivery speed
Firstly, the speed is about creating repeatable processes that push agency and decision making to the delivering teams vs. escalating those same issues and decisions to a governing body like an executive committee, a product manager, a middle manager, etc. And while the context is software, the underlying systems apply to any initiative where a large number of cross-functional teams have to coordinate to deliver something together.
1. Define the goal narrowly enough to rule things out: Large programs often suffer from goals that are too broad and can be open to interpretation. Anthropic's approach: name the specific outcome, the specific stakeholder it serves, and the specific constraint it removes. A goal that rules out most approaches is more useful than one that welcomes all of them. For program teams, this means resisting the urge to align on big objectives and instead forcing agreement on what's out of scope.
2. Deliver in stages, labeled honestly Anthropic calls early releases "Research Preview". This is a signal to everyone involved that this is a pilot, not a finished commitment. For complex programs, this translates to phased delivery with explicit "we're still learning" framing. It lowers the internal bar to move and prevents teams from over-engineering before they get actual feedback from their audience
3. Create a single trigger point for downstream work: Anthropic uses a dedicated Slack channel (their "Evergreen Launch Room") where one team posts when their piece is ready. That single post is enough for every other team to activate. This is a clearly defined handoff trigger. No status meetings, no check-ins with individual teams, no emails announcing that we are moving to the next phase. This is because the roles, responsibilities, and objectives for each functional team have been clearly outlined and are understood from the start.
4. The coordinator owns the handoff process, not the work: The program lead's job isn't doing the work of each team. It's defining when each team gets pulled in and what they need to deliver. When those expectations are set upfront, the "we didn't loop them in on time" delays and finger-pointing when the ball is dropped disappear. This is the difference between a program lead who is chasing status updates and one who is architecting a flow.
5. Shared metrics as a substitute for constant alignment: Anthropic reviews KPIs with the whole team every week. This is not for reporting purposes, it’s so every person understands what matters and why. When teams share a common view of what's working and what isn't, they make better working-group decisions without waiting for central approval. For large programs, a weekly cross-functional metrics review can replace a significant amount of escalation and coordination overhead.
6. A living principles doc instead of per-workstream documentation: Rather than producing a governing doc for team or meeting, Anthropic maintains a single "team principles" document: the key stakeholders, the acceptable trade-offs, and what good looks like. Anyone can consult it and move forward. For program teams, this is a lean version of a program charter — updated as conditions change, used daily, not filed away after kickoff.
7. Use the mission as the tiebreaker: When competing priorities stall a program, the fastest way through is a shared answer to: what are we ultimately here to do? Anthropic resolves cross-team conflicts by asking which option serves the company's mission more directly. For integration programs, post-merger initiatives, or large go-to-market launches, a clearly articulated "why" related to the investment thesis does the same work: it makes trade-offs faster and less political.
8. Run ahead of readiness – and I suspect this one can’t be consistently implemented until the prior 7 steps yield the time-savings necessary for a team to always be 2 steps ahead, because that’s a BIG ask – Most programs have a standing start between phases: one closes, then planning for the next begins. This step suggests that you do the stakeholder prep and design work for the next phase while the current one is still running. In an integration, this means designing the 30–60 day operating model while Day 1 is still underway. When the handoff arrives, there's nothing left to set up.
None of this requires a tech company context. What Anthropic has built is a system where the friction to move is lower than the friction to wait. Clarity on the needs of each person, each team and each delivery are hyperclear and the handoff points and mechanisms have been built before the work begins. These eight practices create a system where work can move fast without losing alignment and quality. It’s a system that we can recreate in our own work, one step at a time or all at once, where the glue holding it together is the cohesion between humans. All without a single mention of an AI bot ☺️
