Skip to content

How corporate engagements work

Corporate engagements deliver one of two things — an auditable result on your data, or a custom AI tool your team uses independently — through four phases: Discovery, a scoped pilot, delivery, and handover. Scope is set on your data and your success criteria, not a fixed product catalogue. Each phase below names what we do, what you see, and what comes out.

  1. Phase 01

    Discovery

    • What we ask: the problem you're trying to solve, what data you already have, and the objectives that would matter.
    • What we look at: representative samples of your data — measurements, inspection reports, lab notebooks, historical archives — under NDA where needed.
    • What we look for: whether the data suits AI-assisted analysis, and whether your team can review AI proposals as part of normal R&D work.
    • What comes out: a written problem statement and a first sketch of what a scoped pilot could look like.
    • Typical duration: 1–3 conversations over 2–4 weeks. No fee at this stage.
  2. Phase 02

    Scoped pilot

    • What's in scope: one problem, one dataset, with the objectives you set and clear conditions for stopping the work.
    • Objectives: what counts as a good result — agreed before we start.
    • Stopping the work: if it turns out this isn't the right problem to bring to AI, we stop. Better to find out early than to grind on something that won't deliver.
    • What you get: either a finished analysis on your data, or a custom AI tool your team uses going forward — decided when scope is locked.
    • Typical duration: 4–12 weeks, depending on dataset size and how detailed the analysis needs to be.
    • What we don't promise: a fixed fee before the initial conversations; outcomes guaranteed in advance; results that bypass your team's review.
    Read the scoped-pilot offer
  3. Phase 03

    Delivery

    • What the AI does: proposes candidate models, extract parameters, computes residuals and uncertainty, flags edge cases and analyses the results based on our and your knowledge combined.
    • What we review: every AI proposal before it lands in a deliverable; every fit before it's reported as a result.
    • What you review: whether the work still matches what we agreed at each milestone; what the final result looks like before handover.
    • Methodology mapping: each step maps to the propose-review loop on the parent methodology page.
    • How often we update you: weekly updates; a mid-project review against the objectives.
    • Where the work needs to run: if it needs to run inside your environment, we adapt — see Data handling.
  4. Phase 04

    Handover

    • What you get back: result data, intermediate artifacts, raw logs, scripts and analysis settings — all in formats your team can re-run.
    • Methodology documentation: the propose-review loop as applied to your engagement, with citations to relevant literature.
    • Option to extend: continue on a new scope, or hand the tool to your team to use independently (if the chosen deliverable was a tool, not a result).
    • What stays with you: everything built on your data.

How AI proposals are reviewed

AI in this work proposes — it doesn't decide. Every proposal is written down in a form your team can audit, and nothing reaches a deliverable without going through the review loop below.

  1. 01ProposeWe use AI and our corrosion knowledge to generate a candidate workflow that can produce a model, an analysis, a parameter set.
  2. 02Reviewwe check the candidate workflow against the data, the engagement scope, and relevant prior literature and modify accordingly.
  3. 03We iterate and refine as neededaccepted workflow proposals are applied to your data; rejected proposals go back to step 1 with feedback recorded.
  4. 04Accept or rejectfinal outcomes are evaluated against the objectives; only results that meet the objectives ship into the deliverable.

Our open demos are working examples of methodology we can apply — they can be modified, extended, or newly built around your data and your workflow. Try EIS Fitting

Confidentiality is a primary engagement constraint — we adapt data handling per project. Read how we handle your data