Skip to content
ITWillRock

Service line 01 / Consulting

Consulting that ends in a decision, including the decision not to use AI.

Book the assessment

Discovery, architecture, roadmaps, and AI readiness from people who install AI pipelines, adapt models, and run software in production. Every engagement ends with a written verdict on where AI belongs in your case and where it does not. You get documents a team can execute, yours to keep, usable with us or with anyone.

01Who this is for

Who this is for

A CTO or head of engineering with a build, migration, or AI decision pending, a board asking "what is our AI plan", and no spare senior capacity to decide well.

An engineering leader whose team already has coding agents on laptops and no pipeline around them: no review gates, no evaluation, no trace, no policy.

A founder or product leader with an idea that needs a model, a device, or both, and nobody senior enough to say whether it will work. Founders who want it built start with the Discovery Sprint; our own products are on the Startups page.

02What we do

What we do

AI readiness and limits assessment
Which processes are candidates, what your data supports, risk classification, build or buy, and a written verdict per candidate naming what an error would cost, what must stay manual, and what AI should not do.
AI-driven SDLC pipeline design
Where coding agents and review gates fit in how your team ships today, what the trace ledger records, what the evaluation suite blocks, how cycle time and defect rate will be measured, and the AI-use policy your auditors will ask for. Written so any team can install it.
Solution architecture
Target architecture, integration map, data flows, model placement (server, device, or none), security and compliance considerations, written as decisions with reasons.
Roadmap
A sequenced plan with options, dependencies, and cost ranges, including the cost to operate models.
Modernization and migration
What to move, how to migrate, what cannot move, where AI shortens the work (code understanding, tests) and where it does not (data semantics, cut-over).
Ad-hoc consultation
A day or a week of a senior architect for a second opinion, including on a vendor's AI claim.

03Entry offer

Entry offer

Two weeks, fixed scope

Architecture and AI Readiness Assessment

Fixed price, quoted after a 30-minute call

  • Week one: interviews; code, infrastructure, and delivery-pipeline walkthrough; data and risk review; one day running two or three real backlog items through a coding agent on a copy of your codebase, in a sandbox in your tenancy or one you approve, under NDA and DPA, so the verdict rests on your code, not a demo.
  • Week two: options; a limits workshop (what AI does here, what it does not, what stays with people); a decision workshop; the written recommendation.

You get

  • A current-state map
  • A target architecture with trade-offs
  • A 12-month roadmap with cost ranges, including the cost to run models
  • An AI opportunity list rated per candidate by value, risk, and data readiness
  • A pipeline blueprint where warranted
  • A written limits verdict naming what AI should do, what it should not, and what must remain a human decision
  • A go or no-go recommendation you own
Book the assessment

04Point of view

Point of view

We live in an agile world, and we still ask you to write a roadmap and a high-level architecture. We work with AI every day, and we still ask you to write down what it must never do. The roadmap keeps the direction steady. The architecture keeps the next sprint from undoing the last one. AI has made the first draft of both cheap. It has not made the decision cheap. Requirements are never final. Plans should survive that, and so should the list of things you decided not to automate.

05How consulting connects to building

How consulting connects to building

Discovery is stage 01 of our lifecycle. If you continue with us, nothing is redone: the pipeline blueprint becomes the AI Pipeline Pilot, the opportunity list becomes the Agent Pilot, and the limits we wrote down become the gates in the build. If you do not, the artifacts are vendor-neutral and any competent team can pick them up. Let us share our bumps and bruises with you, so you do not have to collect your own.

How we work

06Where we have done this

Where we have done this

Financial services

Any financial record can be non-public personal information. We have planned migrations and integrations under that constraint: data classification, encryption, access control, audit, and the question of what stays on-premises. A model is one more system that wants to read that data; we scope its access like any other. Systems we have planned and built were subject to GDPR, PCI DSS, HIPAA, and FDA software requirements.

Medical devices and healthcare

Device software has long cycles, heavy validation, and conservative stacks. We have worked across the software lifecycle of medical devices (software only, not hardware) and we plan with that in mind. An AI component in a device is validated like any other; a model update ships only after re-validation.

07Questions

Questions

Two weeks to a decision.

Start with the assessment. If the answer is "do not build", or "do not put a model here", you will hear it from us first, in writing.

Book the assessment

For sure.