§Book a demo

Thirty minutes. A live audit. Your questions.

We’ll walk through a full audit on synthetic data, then talk about your tool and what evidence you need. If it’s a fit, we’ll discuss the design-partner program.

We use these details only to reply to you. Privacy notice. Prefer email? [email protected]

§What happens next

  1. Within one business day

    A reply from Gideon

    Not an autoresponder. We’ll suggest times for a call.

  2. The call

    Live demo on synthetic data

    See every module on a realistic screener with planted bias.

  3. If it’s a fit

    Mutual NDA, then one export

    Decisions or scores and demographic categories. De-identified data is fine.

  4. About a week later

    Your audit file and walkthrough

    Findings, evidence and the alternatives worth testing.

§Questions

How much does it cost?

Design partners get a full audit and monitoring at no cost for three months, in exchange for regular feedback. After that, pricing is per audited tool per year, agreed with each partner. We’ll share numbers on the first call, and there’s no obligation to continue.

Can Isonomy be our Local Law 144 independent auditor?

Local Law 144 sets strict independence requirements. Isonomy produces the analysis and summary tables in the required format and supports your independent auditor. We’ll discuss on the call whether and how we can act in that role for you.

What data do you need?

One row per person assessed, with the tool’s decision or score and demographic categories. Model inputs, a date, SHAP values and a job-related outcome each unlock more analysis, but none are required to start. See inputs and outputs.

How do you handle our data?

We sign a mutual NDA before any data is shared, work on de-identified data where possible, use it only for your audit, and delete it when the engagement ends or on request. Nothing is shared through this website.

Do we need to integrate anything?

No. Audits start from a file export. Automated re-audits on new batches are part of the design-partner program.

We don’t have SHAP values or outcome data. Is it still useful?

Yes. Impact ratios, intersectional analysis, proxy detection and drift need only decisions, categories and inputs. Explanations and validity-checked alternatives come later, when you have the data.