White-label AI governance for MSPs

Your clients' AI security questionnaires,
answered under your brand.

Customers, partners, insurers and regulators now ask your clients how they use AI. I find out which AI tools they actually use, then deliver the answers and the evidence behind every one of them, with your logo on top.

Pilot: one client, $500. Delivered on time or your money back.
Mapped to ISO/IEC 42001NIST AI RMFEU AI ActSIGCSA AI-CAIQISO/IEC 27001GDPR / UK GDPR
The problem

The AI section of the questionnaire lands on your desk.

A big customer, insurer or partner adds a page of AI questions to its vendor review. Your client forwards it to you, because you're their IT. Nobody on your team is an AI governance specialist, and the deadline is in two weeks.

"We'll get back to you" costs trust. A wrong answer costs more, because once it's signed, someone will check it.

Q1Do you maintain an inventory of AI systems and tools used in your organisation?
Q2Is there a documented and approved policy governing employee use of generative AI?
Q3How do you prevent confidential or customer data from being entered into AI tools?
Q4Can you provide evidence that staff have been trained on acceptable AI use?

Typical questions of the kind found in SIG and CSA AI-CAIQ-style reviews.

What your client gets

Answers they can stand behind, and the proof to back each one.

Results from the sample engagement: a fictional 48-person accounting firm facing a SIG-based security questionnaire with a new AI section.

9
AI tools found
Leadership knew of 3
14/14
Questions answered
Each backed by evidence
48/48
Staff trained
Policy signed, quiz passed
15
Working days
Discovery to sign-off

AI tool inventory

Every AI tool in use, found from sign-ins, browser data and staff interviews, with a decision on each: approved, restricted or banned.

AI use policy and approvals log

A short, plain-English policy your client's leadership actually signs, plus a log of who approved what and when.

Questionnaire answers with evidence

Every answer cross-referenced to a document, record or screenshot, plus a risk register and staff training records.

How it works

Three steps. You stay the face of it.

I work behind your team. Your client only ever deals with you.

Days 1–4

Discover

You share the questionnaire and give me read access, or run my short checklist yourself. I find every AI tool actually in use.

  • Sign-in and app review
  • Short staff survey
  • Shadow AI shortlist
Days 5–9

Decide and document

Your client's leadership approves, restricts or bans each tool. I write the policy, approvals log and risk register.

  • One decision meeting
  • Policy signed
  • Risks recorded
Days 10–15

Answer and evidence

I draft every questionnaire answer, link each one to its evidence, and hand you the pack under your brand.

  • Answers you review first
  • Evidence index
  • Unbranded files
Summary of the Harrow and Vale sample AI governance evidence pack: 9 AI tools found, 14 of 14 answers evidenced, 48 of 48 staff trained, 15 working days
Sample evidence pack

See exactly what you'd hand your client.

A complete worked example for Harrow & Vale LLP, a fictional 48-person accounting firm with offices in Ohio and Manchester, that received a SIG-based security questionnaire with a new AI section.

  • AI tool inventory with nine decisions
  • Approved AI use policy and approvals log
  • All 14 questionnaire answers, each linked to evidence
  • Unbranded, with a space for your MSP's logo

Fictional client, created to show the deliverable. No real client data.

White-label rules

Your clients stay yours. In writing.

Handing a client to an outside partner is a risk to your brand. These commitments go into every agreement.

I never work directly with your clients

I won't contact, pitch or accept work from any client you introduce, during our work together or after it ends.

NDA as standard

Signed before you share anything. Your client's name never appears in my portfolio or posts.

Your brand on everything

Every file arrives unbranded or with your logo. If your client needs to speak to me, I join as part of your team.

Client data stays controlled

I don't put client data into public AI tools. I use the access you grant, nothing more, and delete working files when the engagement ends.

The engineering behind it

Governance written by someone who builds.

I work across cloud, GRC and automation, so the controls I write are ones I know how to implement. Everything below is public, so you can check it yourself.

GRC

Meridian Infotech programme

A full GRC programme for a simulated AI-native IT services firm serving US, EU and Indian banks.

  • 11 AI and cloud risks crosswalked to 9 frameworks
  • Statements of Applicability: all 93 ISO 27001 and 38 ISO 42001 controls
  • AI impact assessments and formal decision records
View the repository →
Cloud

AWS security lab

A real AWS account built to back the Meridian documents with evidence, not just policy.

  • No long-lived access keys; MFA-only human access
  • Data residency enforced by region
  • Evidence fingerprinted and indexed
  • Microsoft Azure Fundamentals (AZ-900)
See the evidence log →
Automation

Controls-first workflows

Self-hosted n8n builds that treat controls as features.

  • AI invoice processor with duplicate and totals checks
  • Billing system with segregation of duties and append-only records
  • LLM pipeline that scans and scores business problems
Browse the builds →
Pricing

Fixed prices, per client. You set your own margin.

You bill your client however you like. I bill you a fixed fee, agreed before work starts.

Founding partner pricing. My first 3 MSP partners lock in these rates for 12 months. Standard prices are shown crossed out.
Start here Pilot
$500 one client
Prepaid. The low-risk way to test working together.
  • One client's AI questionnaire section
  • Evidence pack behind every answer
  • AI tool inventory and use policy
  • Delivered under your brand
Book a pilot call
Starter
from $1,500$2,000 per client, by staff count
A full AI governance baseline, sized to the client.
  • Everything in the pilot
  • Every AI question, up to 2 questionnaires
  • Risk register and approvals log
  • Staff AI training and records
Talk it through
Care
from $250$300 per client / month
Keep it current as tools and questionnaires change.
  • New AI tools reviewed and logged
  • Policy and register updates
  • Next year's questionnaire answered
  • Quarterly evidence refresh
Talk it through
On-time guarantee. If your pilot evidence pack isn't delivered within 15 working days of kickoff, you get a full refund.
Mitchell Swapnil Chatterjee
Who you'd work with

I'm Mitchell. You work with me directly.

I'm a computer science engineer who works across cloud, GRC and automation. I started this service after watching small IT teams get handed AI questions nobody had trained them for, and realising the hard part isn't the paperwork, it's finding out what's really in use and getting a decision on it.

I'm early in building this, which means you get me, not a junior handed your file. I keep a small number of MSP partners so every pack gets proper attention.

Before this I worked in customer support and lead generation, and I've tutored students for two years. Both taught me to explain technical things clearly to people who don't live in them, which is most of what this job is.

B.Tech, Computer Science EngineeringMicrosoft Azure Fundamentals (AZ-900)Based in India, working with UK, EU, US and AU
FAQ

Questions MSPs ask first.

You're in India. How does timing work?

I keep fixed call windows that overlap with UK and EU mornings and afternoons, US East Coast mornings and Australian afternoons. Most of the work happens asynchronously anyway, and you get written updates at each step.

Who signs off the answers?

Your client does, after you've reviewed them. I draft every answer and link it to evidence, but nothing goes back to whoever sent the questionnaire until you and your client are happy with it. Your client's leadership owns the AI decisions; I make sure those decisions are written down and provable.

What do you need from us?

The questionnaire, a named contact at your client, and either read-only access to their sign-in logs or 30 minutes of your engineer's time to run my checklist. Plus one decision meeting with your client's leadership in the second week.

What if the questionnaire is unusual?

Send it to me before we agree anything. I'll tell you within two working days whether it fits the pilot, and quote a fixed price if it doesn't.

Is this legal advice?

No. I map your client's practices to the frameworks and questionnaires they face and document the evidence. Where a question turns on a legal interpretation, I flag it so your client can check with their own counsel.

How do I pay?

By international bank transfer, invoiced in USD. The pilot is prepaid; Starter and Care are invoiced per client.

Next step

Got a client with an AI questionnaire right now?

Book 20 minutes. Bring the questionnaire if you have it, and I'll tell you on the call what it would take to answer it.