Skip to main content
A red marker circles the one missing function in a grid of apps and functions

Current-State Analysis / Feature Definition

The current-state analysis checks every existing app: which functions exist and where a gap remains. Each scored task gets its own feature definition from this before the first build begins. The sequence follows from which gap can be closed and judged the fastest.

Planning With Your Own Data

The plan is built from the company’s own figures and appointments, read straight from scheduling and bookkeeping. For every task the managing director sees which tool does the work, what the build and the monthly running cost are, and how many months it takes to pay for itself.
A consultant in a white business shirt shakes hands with a founder couple in a spacious office with glass walls, the co-founder holding the agreed plan

Acceptance and Extension

The acceptance criteria are fixed before the build begins: who accepts the result, and what tells a correct result apart from a wrong one. After go-live, the customer keeps the cost in view at any time in their own dashboard and extends any task from there within the same project.
Reuse before build
what already works gets reused before anything new starts
Effort and cost
sit on the plan before anything is approved
Better each build
each task ships on its own, and the plan takes in what every build showed

The Build Plan per Task

From Current-State Analysis to Feature Definition 3 steps
01

The Analysis Starts With the Scored Sheet

The current-state analysis takes over the sheet from discovery: the scored tasks, each with hours per month, a suitability total, one leading route and an owner. For every task, the analysis also checks which function exists in the current apps and where the feature definition has to close a gap.
02

The Clearest Feature Definition Goes First

The first task is chosen by how quickly its feature definition stands and can be judged: an owner who recognizes a correct result, systems that are reachable today, and a first build small enough to correct within an afternoon. The saving is weighed next.
03

Feature Definition Follows the Dependencies

Two tasks with the same feature gap and the same connector are planned together, and two with no overlap run in either order. The sequence follows from which tasks need the same permission or the same external decision, with a reason attached to every position.
Cost and Payback per Task 3 steps
04

The Right Tool per Task

Work in Outlook, Excel or Teams runs in Claude Cowork or Copilot Studio. What a standard tool leaves open becomes a Claude skill or an MCP connector, built in a fixed time with a decision at the end and a fixed budget ceiling.
05

Payback Period per Task

Baseline, target and difference are written in one unit: hours per month. Every solution approach becomes directly comparable this way, and the plan records for each task whether it pays back within a year.
06

Build Cost and Running Cost per Solution Approach

A standard tool carries mostly licence cost, a custom build mostly model usage per run and a fixed hosting line. Both figures, the one-time build effort and the monthly running cost, sit on the plan for the chosen approach before anything is approved.
Acceptance Criteria Up Front 2 steps
07

Every Task Goes Into Acceptance on Its Own

Every task is scoped so it ships on its own and is ready for acceptance before the next one starts. What the first build shows goes straight into the next task.
08

Acceptance Criteria Agreed Up Front

Whoever accepts the result today writes the criteria before the build: what a correct result looks like and what happens with a wrong one. That sentence decides when the build stops, and it is what the acceptance is measured against.
Cost Control and Extension in the Dashboard 2 steps
09

Costs Run Live in the Dashboard From Go-Live

Once the build is running, the customer's own dashboard shows the running cost and the new baseline in hours per month, available on any day of the month. The difference from the estimate is the actual saving, visible to the customer directly.
10

Every Build Opens the Next Extension

What the last build showed feeds straight into the next position: which systems ran as expected, which step took longer. An extension builds on the existing task within the same project.

Frequently Asked Questions

How is this different from discovery?
This phase starts where discovery ends. In discovery, systems analysis and a single session with the office count the recurring tasks, measure hours per month, score each one against six criteria and assign a leading route. Planning takes that scored sheet and settles what comes next: which task is built first and which one waits, and what each will cost to run.
How long does the planning itself take?
It is measured in days. The work is the estimating: splitting each task into known work and unfamiliar work, checking that the systems named in its route can actually be reached, and getting the acceptance criteria written by the people who accept the results today. The output is a plan document.
What if the plan turns out to be wrong once the first build starts?
Each task ships on its own for exactly this case. What the first build shows corrects the second position, and the sequence is re-read after every build. The baseline and the acceptance criteria stay fixed, because both were recorded before anything was built and are what the remeasurement is read against.
Is this the same planning as a software project?
It plans office work: the tasks are sequenced against the systems a company runs today, in hours per month and running cost per month. A build on your own codebase is planned on the Needs Analysis & Feasibility page in Agentic Engineering.