Six steps, and you're in control at each one.
Forwargen follows the path a good forward deployed engineer takes: understand first, then diagnose, then propose, then deliver. You check its work at every turn.
- 1
Describe the problem and share what you have
Start an engagement, record a problem in your own words, and say what it costs you and how urgent it is. Then add context: documents (PDF, Word, Excel, Markdown or text), the systems involved, the metrics you track and the people affected.
Forwargen reads and indexes everything you upload so it can search it later, and cite it.
Your part: Write a few sentences and upload what you already have. No preparation needed.
- 2
Answer the questions an experienced engineer would ask
Forwargen interviews you one question at a time. It works through what any diagnosis needs (why now, what goes wrong and what it costs, how the work is done today, the constraints, who decides, and what success looks like), plus the questions for the known patterns your problem resembles. Each question goes after the most important gap that's left.
It checks your documents and earlier answers first, so it doesn't ask what you've already told it. Where an answer is vague ("a lot", "sometimes"), it asks for a number or an example. Where two answers disagree, it asks which is right.
For how the work is done today, it walks you through it one step at a time: who does each step, with what tool, how long it takes, who it goes to next, and what goes wrong there.
Your part: Answer in your own words, skip what you don't know, and wrap up whenever you like. You can stop and come back later. Then send the people who do the work a private link to their own short interview; no account needed. Prefer a form? You can answer the full question list instead.
- 3
Check what it understood
Before diagnosing anything, Forwargen writes back its understanding: your problem in your words, a process map of how the work flows today (a swimlane diagram with one lane per role, and the pain points marked on their steps with what they cost), then the facts it is working from, grouped by topic, plus the gaps it still sees.
Each fact is tagged: backed by a document (with the quote), confirmed by two people, said by you or a colleague, or our inference (please check). Where people disagree, it shows each view side by side instead of picking one. It also catches contradictions, such as a figure in your notes that doesn't match your export.
Your part: Correct, remove or add statements and steps, then confirm. Diagnosis waits until you do.
- 4
See the root causes, with the evidence
Forwargen searches your documents and its industry knowledge and builds a diagnosis: the likely root causes, the evidence behind each one quoted from your material, a confidence level, and the assumptions and open questions that remain.
Citations are checked in code against the passages it actually retrieved. Anything it can't back up is labelled an assumption.
Your part: Confirm the diagnosis, or say what's wrong and run it again.
- 5
Compare the options, including doing nothing
You get two to four ranked options next to a "do nothing" baseline that spells out what happens if nothing changes. Each option says what kind of change it is (a process change, configuration, an integration or a new tool), whether it fits the team you described, its effort and risk, and the prerequisites.
Each option also carries a business case as a range, left blank when your evidence can't support a figure, and a quick win you could start within 30 days.
Your part: Pick the option you want to pursue.
- 6
Get the documents to carry it out
Forwargen produces the documents the chosen option needs, starting with a discovery plan: the problem, current state with the process map, stakeholders, requirements, options, roadmap, business case, risks and next steps. Depending on the option, it adds technical specifications, data mappings, process documents, runbooks and rollout plans.
Each document is scored by a separate AI reviewer against a rubric for that document type. If it falls short, it is revised once automatically. Every change is a new version.
Your part: Review, comment, request changes and approve. Export to Word, PDF or Excel.
Every document gets an AI review.
Before you see a document, Forwargen reviews it with AI against a checklist for that kind of document, and fixes what falls short. Then your team reads it and approves it.
What Forwargen doesn't do
- It doesn't log in to or change your systems. It produces documents; your people, or an implementer you choose, carry them out.
- It doesn't present its own inferences as facts. They're labelled, and you confirm them.
- It doesn't invent figures to make a business case look better.
See it on one of your own problems.
A pilot starts with a single problem and the documents you already have.