What should we build, and is it worth it?
A costed build plan in two weeks, or a written reason not to build.
Two to four weeks of opportunity mapping, target architecture and a costed business case, ending in a build plan you can approve, re-tender or walk away from.
What it is
We spend two to four weeks inside the operation: reading the systems, sitting with the people doing the work, and counting where the hours and the margin actually go. It ends in three artefacts you own — a ranked opportunity list with a rupee figure against each line, a target architecture drawn against the stack you already run, and a build plan carrying a price and a date. If the numbers do not support building, the deliverable says so, and you keep the artefacts either way.
Who it is for
If none of these sound like your situation, this is probably the wrong arc — and we would rather tell you that than sell you a discovery.
- You have an approved AI budget for the year and nothing written down about what it buys.
- Three vendors quoted the same platform at prices an order of magnitude apart and you cannot tell which quote is honest.
- Two departments each want an agent built and you can fund one of them this quarter.
- The last vendor shipped something that works, nobody uses it, and you need the reason before you spend again.
How the work is shaped
Repeatable shapes, not bespoke proposals. Each one has been run before and has a duration we hold to.
- 01Opportunity Map · 2 weeksSix to ten interviews across the teams that own the work, a read of the systems behind them, and a ranked list of candidate builds with a cost and a payback figure on each line.
- 02Architecture Review · 3 weeksWe draw the target architecture against your current stack — data, identity, integrations, tenancy — and mark what has to change before anything is built on top of it.
- 03Build Plan · 4 weeksThe chosen opportunity becomes a scoped plan: slices, sequence, team shape, the integration risks named, and a fixed price for the first slice.
The JJ Tech Opportunity Map
How the work is actually run
Two weeks, in this order: interviews, a read of the systems behind them, then one day in a room — because a room that has not been measured first ranks opinions rather than opportunities.
- 01Where the money goesWhich parts of the operation actually carry the cost, and can finance put a figure on them?Leaves behind: Two or three domains, each with an annual rupee figure and the name of the person who owns it.
- 02Handoffs, timedWhere does the work wait — between which two systems, for how long, and how often?Leaves behind: A timed handoff map with the real constraint marked, which is rarely the stage everyone names.
- 03CandidatesWhat could a model or an agent change here that a rule, a report or a fixed integration could not?Leaves behind: A candidate list, each one written as the smallest slice that would be worth shipping alone.
- 04ArithmeticWhat does each candidate cost to build, to run and to supervise, and in which month does it pay back?Leaves behind: A payback line per candidate with its assumptions exposed — including the ones that would make it fail.
- 05First sliceIf you fund exactly one thing this quarter, which one, and what does it cost?Leaves behind: A build plan for the first slice at a fixed price, and a written reason for every candidate not chosen.
What gets ranked, and on what
Three tests, applied to every candidate in front of the people who will have to live with the answer. A ranking whose criteria are not published is an opinion with a table drawn around it.
- Impact
- Name the number that moves and the person whose week gets better. A candidate with no nameable beneficiary is parked, not ranked.
- Feasibility
- Could a working slice be in someone's hands in six weeks — judged with an engineer in the room? Feasibility scored without one is optimism with a number against it.
- Data readiness
- Does the data exist today, resolvable to one identity, with someone who can grant access this week? Where it does not, that reconciliation is the first project and the model is the second.
What sits inside it
AI opportunity mapping
Where the hours and the margin go, ranked by what an agent or a platform could actually change.
Scoping
The chosen opportunity written down as slices, sequence and acceptance criteria a vendor can be held to.
Architecture
Target architecture against your stack: data, identity, tenancy, integrations, and what must change first.
Business case
A cost model with its assumptions exposed — build, run, model spend, and the month it pays back.
What it has produced
Every number here comes from work on this page. Follow it to the story and check it.
- of annual margin leakage found across 11 lanes
- ₹1.4 Crof annual margin leakage found across 11 lanesSee the story
- from kickoff to a costed build plan
- 14 daysfrom kickoff to a costed build planSee the story
- artefacts you own and can take to another vendor
- 3artefacts you own and can take to another vendor
- of price bands published before the first meeting
- 100%of price bands published before the first meeting
Is Envision what you need?
Tell us what is slow and what it is costing. If the answer is a different arc, or no arc at all, we will say so before anyone writes a proposal.
Map the opportunity