Strategic Initiatives

Strategic initiatives are the significant efforts your organization decides to pursue - the output of strategy work, captured as a concrete, prioritized list.

Recording and reviewing initiatives

The everyday way to work with initiatives is to talk to the AI Strategist: use a strategy session to explore where the business should focus, then capture the outcome as initiatives right in the conversation. The Strategist can also review your existing list with you and help you refine or reprioritize it - it confirms every change before recording it. For the full story of what the Strategist does and why, see The AI Strategist on our website.

The initiatives list

For administrators, the Strategic Initiatives page (under Administration → AI Strategist) shows every initiative with its owner and status, ordered by priority. The priority order is explicit and shared: everyone working from it sees the same ranked list, so there’s no ambiguity about what matters most.

Prioritizing

On the initiatives page, drag an initiative up or down the list to change its priority - or simply tell the Strategist how the priorities have changed. The new order is saved immediately and everyone sees the same ranking.

How your initiatives shape what the Strategist tells you

Recording initiatives is how you point the Strategist’s attention. Once they are on the list, it connects each finding its analysts make to the initiative that finding advances or threatens, so what you see is framed by what you said matters:

Your AI Daily Briefing leads with the priorities you set: findings that bear on a stated initiative come first, and each one names the initiative it serves.
The briefing reports progress toward what each initiative is for. If your description names something as only a means to an end, such as keeping records complete so an analysis can run, the briefing covers it in a short progress line rather than making it the headline.
Decisions waiting on you are labeled with the initiative they affect, and ones that advance a stated priority rise toward the top of your queue.
When an initiative says you keep records in order to learn something from them (which parts to stock, which customers pay late), the Strategist starts an Ongoing Analysis on that process itself and tells you what it is working on and how often. There is nothing to approve; ask it to reword the question or stop.
Silence is reported too: when nothing is moving on one of your initiatives, the Strategist says so plainly rather than letting it drift out of sight. That includes an initiative with plenty of activity around it but nothing yet toward its goal.

When priorities change, update the list (or just tell the Strategist in conversation). It reorganizes around the new priorities from the next briefing on.

How the Strategist puts an initiative to work

Once an initiative is on the list, the Strategist looks every day for the next piece of work that would move it forward and brings it to you as a suggestion. Nothing it suggests happens until you approve it. It suggests three kinds of work:

Run a process you already have - for example, start your vendor vetting process for the new office, with the region already filled in. If you built a process that fits but have never used it, the Strategist says so.
Design a new process - when nothing you have covers the work. The Strategist designs the steps, outcomes and fields, and chooses who does each step: a team when one fits the work, or one person when the step is clearly theirs. It never picks a group that holds most of your company, such as an everyone group, or anyone whose access is read only. The first step always goes to someone it named, and so does the step after any step an AI Agent does. When a step is information work an AI Agent could do, such as summarizing the notes from a site visit, the design can give that step its own AI Agent, and the suggestion shows the agent’s instructions in full. Approving creates the process, with any AI Agent, and starts its first run. The people it named get their tasks as the run reaches their steps, with the usual task email unless they have turned task emails off. A step an AI Agent does sends no task email, because the agent does the work. If the agent can’t finish it, the task stays open for whoever the step is assigned to.
Have an AI Agent do a step - when a step in one of your processes is information work an AI Agent could do, such as researching the vendor named on each record.

The Strategist never changes one of your processes without your approval. Adding an AI Agent to a step, like any other change to a process, happens only when you approve the suggestion or ask for it in a conversation.

If someone a suggestion names has been removed or can no longer do tasks, approving it is refused and nothing is created. A suggestion for an initiative that is no longer active can’t be approved either: set the initiative back to active first.

To have a step of a new process go to someone else, approve it and then tell the Strategist who should do that step. It changes the process for every task from then on. If the step’s task is already waiting, ask it to move that task too.

Give an initiative a finish line

An initiative with a target, such as “cut quote turnaround to 2 days by March 31”, can have a finish line: the Strategist measures the target itself, every day, on one of your processes, and tells you where it stands. The numbers come straight from that process’s runs and are worked out the same way every time, so they are never an estimate or a guess.

What it can measure

Cycle time - how long a run of the process takes from start to finish. Lower is better.
Runs abandoned before finishing - the share of started runs that were canceled instead of completed. Lower is better.
Runs hitting a problem - the share of finished runs where someone chose a response whose AI Strategist Signal is unfavorable. Lower is better.
Steps finishing late - the share of steps with a due date that were completed after it. Lower is better.
Completed runs - how many runs of the process finished since you recorded the initiative, counted toward a total. Higher is better.

Each initiative has one finish line. The target has to point the better way for its measure: the Strategist won’t measure “increase cycle time”, and tells you why instead.

Set it up

In a strategy session, tell the Strategist which process to measure and how: “Measure this as the cycle time of our Quote process.”
It reads back exactly what it will measure, for example “Quote cycle time, in days”, and waits for your yes before recording anything. When it records it, it also says the target it measures against, such as “2 days or less”, so a share typed as 0.05 when you meant 5% is easy to spot.
Its first reading comes with its next daily check.

The process has to be one you can see. To measure an initiative differently, or to stop measuring it, tell the Strategist the same way. A new measure gets its own starting point.

If one of your initiatives has a target that nothing measures yet, the Strategist may pick a process and a measure for it on its own. When it does, it tells you exactly what it chose, and you can correct it or ask it to stop in conversation. It never picks one again for an initiative you stopped measuring. A target about money, such as revenue or profit, is never measured by counting runs, whether the Strategist picks the measure or you do.

Ask where it stands

Ask the Strategist how an initiative is doing, and it tells you:

where the number stood when you handed the initiative to it. That starting point is set once and never moves;
where the number is today;
when your target has a date, whether you are ahead of, on, or behind the straight line from where you started to your target by that date. For completed runs it also says where the current pace will land by the date.

Your AI Daily Briefing says where each measured initiative stands too, and comes back to it when that changes: falling behind, catching up, or reaching the target.

A reading needs enough recent runs: at least 5 finished in the last 28 days for cycle time, and at least 20 for a share. When a process runs rarely, the Strategist reads its most recent runs from the last six months instead and says so. When even that is not enough, it tells you how many runs finished and waits for more.

A share counts every run, however busy the process is. Cycle time reads no more than a day’s most recent 2,000 runs. On a process that busy, the cycle-time reading says how far back those runs go.

When the target is met

A finish line is met when the reading has stayed on the target side every day for two weeks, or sooner on a busy process: once the days it has held contain 2,000 runs (2,000 steps for late steps). A completed-runs target is met the day the count reaches it.

Then the Strategist asks you to mark the initiative achieved. The request waits in your Decision Queue and comes up with your other decisions in your strategy session. Answer it in the conversation, or on the request itself:

Approve & mark achieved - the initiative is marked achieved, and the Strategist stops measuring it.
Reject Proposal - the initiative stays active and the Strategist keeps measuring. It asks again only after the target has held for another two weeks, or across another 2,000 runs if that comes first (for completed runs, two weeks later).

If the number slips, or the initiative is paused or can no longer be measured, before anyone answers, the Strategist takes its request down, and the Decision Journal records why. It never marks an initiative achieved on its own. Answering the request needs the Guide AI Strategist permission.

If the number already met your target on the day you recorded the initiative, the Strategist tells you so instead of asking, because there was no gap to close. If it later slips past the target, the Strategist tells you it has slipped and that you’re behind. An initiative measured without a target number has no finish to reach: the Strategist tells you which way the number is moving.

When an initiative changes

Paused - the Strategist stops measuring until the initiative is active again.
Achieved or Abandoned - the finish line closes with its last reading, whether the Strategist asked or you changed the status yourself.
Set back to active after being achieved - the Strategist treats that as “not yet”. It measures again, and asks only once the target has held for two weeks from then, or across 2,000 runs if that comes first.
Deleted - the finish line goes with it.
The process is retired or deleted - the Strategist tells you it can no longer measure the initiative, so you can pick another process.