Next live sessionThu, Nov 5 · 11 am ET · 8 am PT · 16:00 UKStarts soon
Know which questions AI can answer from your data today.
Data readiness for AI is a free, one-hour live workshop. Before an AI agent answers from a company's data, we run seven checks on it. You'll see each one, and how each part of the business gets a verdict: ask now, ready in weeks, or data work first.
What you'll get
- Which questions AI can answer from your data today
- The seven checks, and how to score yourself
- What to fix first, and what to leave
11 am ET · 8 am PT · 16:00 UK · online, one hour, free
Teams we've built for
- Comic Relief
AWS Premier Tier Services Partner, one of about 300 among more than 130,000 AWS partners.
Point an AI agent at messy data and it repeats the mess, faster.
When an AI project stalls, the cause is usually the data underneath the model: a word with three meanings, a customer with five records, or a field renamed on a Tuesday that makes Wednesday's answer read $0 without a sound.
None of it shows up in a demo, because a demo runs on clean sample data. It shows up in week six, when the finance lead holds the AI agent's number next to their own and they don't match.
Data readiness for AI means finding those gaps before you build, one part of the business at a time, so you know which questions an AI agent can answer today, and which need work first.
What you'll take away from the hour
- 01
One meaning per number
Why nobody's figure is wrong: they're three different numbers sharing one name, and how to give each one owner.
- 02
A score you can trust
Meaning, source, quality, access, change, proof and history, each scored for each part of the business.
- 03
Where to start
The first area to make ready, and the questions with known answers that prove it's ready.
Seven checks on your data, scored for each part of the business
Each check is scored for each area (sales, finance, operations): 0 nobody knows, 1 it lives in people's heads, 2 it's written down, 3 it's enforced and tested. These are the questions a CTO would ask before letting an AI agent answer for the company.
- 1
Meaning
Does each word that matters have one meaning, written down, with an owner? We look for a glossary, metric definitions, a semantic layer or a catalog with business descriptions. Without it, the AI agent picks a table and sounds sure.
- 2
Source
Is there one place each number lives, and can systems be joined on a key that's the same everywhere: one customer ID across the CRM, billing and support? Five records for one person give five answers.
- 3
Quality
Are the fields that matter filled in, consistent and fresh? We count how complete the key fields are, how many spellings one thing has, the duplicates and how late data arrives. That matters because an AI agent reads a blank and a zero the same way.
- 4
Access
Does an AI agent see only what the person asking may see? We look for row- and column-level rules in the data platform, with the asker's identity carried on every query. A rule written into a prompt can be argued away, and a filter in the data platform can't be.
- 5
Change
When a field is renamed, does anything notice before an answer goes wrong? We look for schema change detection, data contracts and a pipeline that pauses and tells its owner. This is the silent $0.
- 6
Proof
Are there questions with known answers to test any AI agent against, and does every answer show its working: the query, the sources, the time? Without them, nobody can say the AI agent is right, or that it got better.
- 7
History and documents
Is there enough history for the questions people ask, and how much lives only in spreadsheets, PDFs or one person's head? A trend can't be read without history, and nobody can query what one person remembers.
We give each area one of three verdicts. Ask now means an AI agent can answer those questions today, ready in weeks means it can once the meanings are written down, and data work first means joining, cleaning or keeping history comes first, as a project of its own.
What we bring
Once your data is ready, this is what your AI agents read it through.
Axiom
A semantic layer that gives every AI agent one meaning of revenue, customer and churn, worked out from the agreed definition instead of guessed, with each answer showing its working.
Runs in your AWS account; any AI agent reaches it over MCP.
One hour, run by the engineers who do the work
See it
Axiom, on an example business, is asked “What was September's revenue?” It asks which revenue instead of guessing, answers, and shows its receipt: the meaning and who agreed it, where it lives, the query and the time. Then a field is renamed, and the pipeline stops and tells its owner instead of answering $0.
How it's scored
The seven checks, one at a time, on the same example business: what we look at, what a 0 and a 3 look like, and the verdict each area gets.
Your questions
Bring the number your teams argue about. We'll take as many as the hour allows, and tell you honestly which ones an AI agent could answer today.
The private workshop: half a day, with three to six of your people
The same checks, run on your own data. The moment people remember comes after the break: your own questions are asked of a sample of your data and scored against the known answers; then the meanings agreed that morning are written down, and the same questions are asked again. The change in the score, before and after, is the point of the day.
Who comes
The owner or operations lead, who decides; the finance lead, who owns what “revenue” means; one person from each team that uses the number; the person who builds the reports or keeps the systems.
What to bring
The number your teams argue about, and 10 to 20 of the questions people ask most, with known answers. Last quarter's board pack is the easiest source.
Your data
Nothing personal has to leave your systems: read-only access to two or three tables, an export with personal data taken out, or nothing at all, in which case we build a look-alike from what you describe.
What you leave with
- The words that matter, each with one meaning, an owner and where it lives: the first page of your own semantic layer
- The scorecard and a verdict for each area of the business
- Your own questions, scored before and after the meanings were written down
- A draft data model for the first area to make ready
- What to fix yourselves first, if it isn't worth bringing us in yet
What clients say about working with us
Armakuni exemplify best practice. Our team is more confident, more effective and more knowledgeable thanks to our engagement with them.
Thanks to Armakuni's efforts, the client saw improved operational efficiency and reduced manual workload.
Armakuni goes above and beyond to deliver and copes with mistakes instead of hiding them.
- Under 0.25%of AWS partners are Premier Tier. We're one of about 300.
- 500+AWS customer launches
- 1,300+engineers across the UK, US, UAE, India and Canada
- 70%of the proofs of concept we scope reach production
Before you sign up
Is it free?
Yes. The hour and the half day on your own data cost nothing; AWS has programmes that can fund the work after it, and we'll say if one fits.
What if our data isn't ready?
Most isn't, in at least one area. You'll hear which, and what to fix yourselves first, before anyone suggests a project.
Do we have to share our data?
Not for the live session. For a private one you choose: read-only access to two or three tables, an export with personal data taken out, or nothing, and we build a look-alike from what you describe.
Who should come?
Whoever owns the number, whoever builds the reports, and anyone about to put an AI agent in front of the company's data.
What happens after?
If it's worth doing, a fixed-scope first look of one to two weeks, then first answers live in about 90 days, with a go-or-stop at week 4. If it isn't, you keep the scorecard and the plan.
Data readiness for AI
Thursday, November 5, 11 am ET · 8 am PT · 16:00 UK. Starts soon.
Save my seat