Guides
Decide beforeyou spend.
Practical guides on cost, scope, hiring, innovation, R&D, and AI — so you can buy, kill, or build with your eyes open.
01
Hire, cost & control
Budgets, briefs, timelines, and how to pick a partner without losing the repo.
01
How to hire an innovation agency
A hiring sequence: name the job, ask six questions, compare two scopes, walk away from a pitch. Built for companies that need a system they can keep.
Read the guide
How to write an innovation brief
One page, not a deck: the stuck job, the unknown, what you already tried, the kill rule, and what you will not fund.
- 03
How to read a statement of work
The lines that have to exist: what ships, what is out, when the card runs, who owns the repo, and how you cancel. If they are missing, it is a pitch.
- 04
How to budget for custom software
Price the job, not a platform. Site vs workflow vs product, five-year cost, and what a $750 Spark is actually for.
- 05
How to choose custom vs off-the-shelf
A decision test: buy when the job is common, build when the workaround is the company, sequence a migration so you are not betting the week.
02
Innovation & R&D
What the words mean, what a studio ships, and when a lab is just the backlog.
01
How to run product innovation
How to tell if a new product is real, who it is for, and when to stop before you spend money building it.
Read the guide
- 02
How to run productivity innovation
How to make one job inside the company take less time, and how to prove it before you buy another tool.
- 03
How to find the job worth funding
Walk the work. Score waits, rework, and hero jobs. Fund a constraint with a user — not a whiteboard of ideas.
- 04
How to test an idea before you build
A test is a user, a thin lie, and a number. If you cannot fake it this week, you are not ready to spend on the real thing.
- 05
How to stand up an R&D pod
Do not hire a chief of innovation first. One unknown, two or three people, a kill rule on day one, and a door into the real work.
- 06
How to de-risk R&D
How SINLE de-risks research and development: name the stuck job, run a patent search and a thin Tuesday test, write a kill rule, then spend. A bigger lab is not the method.
- 07
How to run a patent search before you fund the build
The first-pass patent search SINLE runs before a build: the stuck job in plain words, synonyms, claims, a one-page note. Find the fence before Spark becomes Invent.
- 08
How to monitor patents while an R&D bet is live
How SINLE keeps a live R&D bet from walking into a new fence: the same search, monthly, scored in one line, next to Pod — not an industry dashboard.
- 09
How to kill a project that is not working
Use the stop rule you wrote on day one. A clean kill is a result. A zombie project is how next month disappears.
- 010
How to measure whether the new way stuck
One outcome metric, one adoption metric, a date. Ideas and patents measure activity. Friday measures innovation.
03
AI in production
Clerks, locks, and when a model is the wrong tool.
- 01How to pick the first AI jobCircle work that is frequent, textual, and cheap to get wrong. Publish the truth first. Then a clerk — not a strategy.Read the guide
- 02How to ship an agent with a human lockThe lock is the product. The model is the clerk. Auth, keys, evals, a kill switch, and a person before money or a send.
- 03How to tell AI is the wrong toolA filter you can run in a meeting: undo-by-Friday, named job, owned source of truth. Fail any line and you are buying an incident.
04
Build & ship
From prototype to a URL the company can keep — spreadsheets, MVPs, products.
01
How to replace a spreadsheet with software you own
Map the real columns, agree who confirms a row, propose-then-confirm, then the sheet becomes an export — not the company.
Read the guide
- 02
How to go from prototype to production
Production means auth, real data, a kill switch, and a URL the company can send. The path in between, without a two-year programme.
- 03
How to scope an MVP
Minimum means one core workflow in a real user’s hands. It does not mean half of everything. A 90-day shape you can actually calendar.
Next
Need an answer for your job?
Send the stuck process in one sentence. We reply with a scope, a Spark, or a written no.