Three families, six solution types
Order matters
The changes build on each other. Remove waste, then reengineer what’s left, then automate the simpler route — fixed rules first, then agents for the variation that rules can’t handle — and then change the foundation and the organization around it. Automating a step you could have removed just makes the waste faster. Then zoom out. Order to Cash alone is several connected processes in one value stream, and a real company runs hundreds of processes with hundreds or thousands of people. Read the solution set across the value stream, not one process at a time.How to choose the mix
1
Start from the root causes
Use Advisor to find the largest causes of the gap in your Work Brain (Process Index), not a favourite solution. Size the causes together: fixing one usually exposes the next. See Strategic analysis with Advisor.
2
Remove and reengineer first
For each cause, ask what work can come out and what can be rearranged before anything is automated. These changes are often the fastest and the cheapest to run.
3
Pick the smallest automation that works
Choose the smallest shape that produces the outcome and passes its controls. Use rule-based automation when the process is stable and every step must be explainable. Use an agent when the inputs vary too much for a fixed path. Every agent option names its non-agent alternative, so you can compare the two.
4
Name the enterprise changes
Note the system, data and organizational changes the outcome depends on, and who owns them. Your teams deliver these; Within makes the dependencies visible.
5
Agree the future state, owners and investment
The solution set is done when the future state, the mix of changes and the investment are approved, with an owner for each change.
Is an agent ready to build?
Grade each agent candidate before it goes on the roadmap, and check system access here rather than after a build.
Then pick how much the agent may do on its own:
- Process redesign only. No agent yet: the change is to the process.
- Draft-only. The agent prepares the work and a person approves it.
- Full autonomy. The agent acts on its own. Keep a person’s approval for anything irreversible, regulated or expensive to correct, and reduce review only when testing and production evidence support it.
Who decides what
Doing it in Within
Advisor does the analysis across your Work Brain. Two prompts to start from:Solution mix for each root cause
The non-agent alternative
Where to go next
Build your AI roadmap
Set the objective and sequence the changes.
Execute your improvements
How each solution type runs through delivery.
Redesign the process
Remove waste and reengineer before you automate.
Build an agent with the Within ADK
Specify, build and validate an agent from observed work.
Need help? Use the in-app chat — click the chat bubble in the bottom-right corner of Within (staffed 24×5) — or email support@within.ai.

