Book a conversation
Menu

Evident

AI that works in your business and demonstrably holds up.

Anyone can switch on a chatbot. Trusting it takes agreements on what the AI may do, access that is right and a measurement that shows it. Evident sets up AI on your processes, with a piece of evidence at every step.

See what is possiblePlan a conversation
  1. 01UnderstandThe hours and the data are measured before anything changes.Evidence A measurement of the hours and an overview of the data, with the first agreements on which data may go in.
  2. 02Set upAI starts working on one or two processes, and it is decided who may do what.Evidence A working set-up on one or two processes, with who may do what on record.
  3. 03EmbedThe team learns to work with it and the choices are embedded.Evidence Work instructions and an owner per process in the organisation.
  4. 04Hand overThe team carries on by itself and the hours are measured again.Evidence A handover and a follow-up measurement of the hours.
The four steps in the Method →

What is possible now

In all six, the AI does the groundwork, a person decides and a piece of evidence remains.

An agent in your business software

From application email to onboarding in AFAS: the agent does the groundwork, your people decide.

Illustration on demo data

  1. 07:00 New email: application Planning officer, J. de Vries (cv.pdf)
  2. Checked against the profile: 4 of 5 requirements, 6 years in planning
  3. Proposal: invite. For recruitment to decide.
  4. Recruitment approved: candidate is in AFAS, application process started
  5. Log: email 07:00, proposal 07:01, recruitment decision 08:14
Read more →

A meeting that ends in actions

Recording, summary and actions straight into the systems already in place.

Illustration on demo data

  1. Monday site meeting: 42 minutes recorded
  2. Transcribed and summarised
  3. 3 actions, each with owner and date
  4. Actions added to the planning, summary in Teams
Read more →

A knowledge base that answers with a source

Manuals, contracts and procedures searchable in plain language, every answer with a footnote.

Illustration on demo data

  1. What is the warranty period on a heat pump?
  2. Searched: 214 documents
  3. Five years on the compressor, two years on the other parts.
  4. Source: Warranty terms 2026, section 4.2, page 7
Read more →

An agent for customers or at the stand

Answers questions on the site or at a fair and passes on the right requests.

Illustration on demo data

  1. Visitor: can the programme be followed part time?
  2. Yes, two evenings a week, starting in February and in September.
  3. Follow-up asked: name and preferred start date
  4. Request passed to the study adviser
Read more →

An access and security scan

Who can do what in your systems, what is too broad and what should be closed.

Illustration on demo data

  1. Scan started: Microsoft 365 and accounting, 120 accounts
  2. Roles compared with job functions
  3. 3 findings: 2 old accounts active, 1 case of creating suppliers and paying them in one hand
  4. Report ready, with a proposal per finding
Read more →

A simulation or training

An exercise in which your team decides and the AI plays back the consequence.

Illustration on demo data

  1. Alert 16:42: unknown login on the payroll system
  2. The team chooses: report first or investigate first?
  3. Chosen: investigate first. Consequence: the 72 hour reporting window is running.
  4. Debrief prepared, with three lessons
Read more →

All capabilities

Demonstrable: what it delivers is measured, and what the AI did can be read back.

What it delivers: hours before and after

Before the start there is a baseline of the hours. Afterwards the same measurement is repeated. The difference is what it delivered.

Worked example

2,250hours per year

45 working weeks →50 employees →1 cell = 1 hour

1 hour per employee per week × 50 employees × 45 working weeks

Source: worked example from the Kickstart proposal. Not a client result and not a promise of time saved.

Measure first, then set up, then measure again.

What it did: work you can point to

Every step leaves something you can point to: a log, a test list, a follow-up measurement, a handover file. A practical case shows what that looks like.

Case, anonymous

An organisation without a technical background

Processes analysed and formalised, Claude rolled out with its own skills and a knowledge base, and the house style fully renewed. According to the organisation the work goes more smoothly; the measurement in hours is to follow.

Read the whole case →

Choose your entry

Two entries, the same four steps.

For SMEs

Evident AI Kickstart

In four to six weeks, AI is set up on one or two processes, with the hours measured before and after. Then your team can carry on by itself.

See Evident AI Kickstart →

With a control environment

Enterprise

AI on the control and reporting processes, with policy, logging and handover on record. The same four steps.

See Enterprise →

Explainers on AI that demonstrably holds up.

Explainer

Why Evident: AI that works and demonstrably holds up

Anyone can switch on a chatbot. Trusting it takes agreements, and something in hand after every step.

Contact

Where are you losing hours?

Send an email with your question and the process it concerns. A request starts with a conversation.

contact@evidentai.nl

Contact