Enterprise / Pilot
Batch Estate Assessment + Replay Proof
Two weeks. Bounded scope. Artifacts, not slides.
A pilot exists to answer one question with evidence: can a meaningful slice of this estate be replaced safely, and how do we know? We take a bounded slice, scan it, convert it, replay your own run history against it, and hand you the bundle — including any divergences, which are the valuable part.
- Duration
- Two weeks
- Scope
- Up to 500 jobs
- Price
- Fixed, quoted up front
- Buy via
- AWS Marketplace or direct
What we need from you
| Input | Typically from | Notes |
|---|---|---|
| Estate export | autorep -J PREFIX% -q, Control-M DEFTABLE XML, BAW .twx, or Pega export | The definitions, not the runtime. Read-only. |
| Run history | Scheduler reporting export, 10–30 days | Job, date, status; start and end times sharpen the ordering checks. |
| One operator | Whoever knows the chain | Two or three conversations, not a workshop series. |
| Nothing else | — | No production access, no credentials, no network path into your estate. |
The tooling is stdlib-only Python with zero network I/O. It can run entirely inside your boundary — on your laptop, your jump host, or your VPC — with us never touching the data. That is the normal way to do this, not the special case. See Trust.
The two weeks
| Stage | What happens | You get |
|---|---|---|
| Days 1–2 | Scan the slice; inventory processes, classify tasks, score agent-readiness, surface the human gates and the undocumented chains. | Estate report (HTML + JSON) |
| Days 3–5 | Convert the chain into an event-sourced graph. Reconcile the shape with your operator — this is where "nobody remembers why that dependency exists" gets written down. | Graph review notes |
| Days 6–9 | Replay the history window day by day. Investigate every divergence: is it our model, or is it a genuine surprise in your estate? | Proof bundle, per-day verdicts |
| Days 10–12 | Write the migration plan against what the proof showed: what moves first, what needs discovery, what stays human. | Migration plan |
| Day 13–14 | Walk the artifacts through with your risk and operations people. No slideware — we read the bundle together. | Readout |
What "success" means — stated before we start
A pilot that comes back PASS tells you a chain is ready to move. A pilot that comes back DIVERGED is not a failed pilot — it has just told you, cheaply and before cutover, that your estate does something nobody documented: a force-start convention, a calendar nobody modeled, a dependency that exists only in someone's head. Both outcomes are worth the fee. We will not quietly tune a model until it passes; that would be manufacturing the answer you paid us to find.
Nor will we claim more than the artifact supports. A proof over a partial model reports its coverage gaps and refuses to pass, by design.
After the pilot
- Nothing. You keep the artifacts and the plan; they are yours, they are self-contained, and they are readable without our software. That is a legitimate outcome and it happens.
- More proofs. Your team runs
handeware-proveacross the rest of the estate under a product license. - The runtime. Chains that proved out move onto the Handeware Engine under a multi-year agreement through AWS Marketplace, drawing down committed spend.
Start the conversation with your estate, not a form
Tell us which scheduler, roughly how many jobs, and whether you can export run history. That is enough for us to say whether a pilot makes sense — and to quote it.
support@handeware.com · subject: “Replay proof pilot”