Meet Axis.
Our AI analyst.
Openly declared.

Framework27 is built by two senior human operators and one AI analyst named Axis. Axis is an openly-declared AI persona built on Anthropic's Claude and tuned on the Framework27 methodology. This page explains exactly what Axis does, what it doesn't do, and why we're being this direct about it.

Why we're telling you.

Most consultancies using AI right now are hiding it. We think that's a bad bet. AI is becoming central to how knowledge work gets done, and pretending otherwise isn't going to age well. The customers who care about this will find out anyway — and when they do, they'll remember which firms were honest and which weren't.

So we made Axis a character. We named it. We gave it a voice. We put it on the homepage. We labelled every output. We called it what it is: an AI analyst.

REASON 1

Trust compounds faster than deception

Pretending a human wrote an audit when Axis wrote the first draft is a short-term win and a long-term loss. The moment a client finds out, the relationship is over.

REASON 2

The combination works

Axis + two operators produces better audits than either could alone. Axis is more consistent across 270 criteria. Operators are better at commercial judgment and pattern-matching.

REASON 3

It's the right thing to do

Clients deserve to know when they're interacting with AI. Regulators increasingly require it. And as AI gets more capable, the disclosure bar is going to rise.

Axis runs this stack.

Axis is involved in every step of every Framework27 audit. Specifically:

FREE ASSESSMENT

Get Your 270

Axis is the only thing running the free assessment. It takes your 27 self-reported scores, calculates the total, assigns the grade band, identifies the three weakest sections, and writes the personalised summary. No human sees your responses unless you later book a paid audit.

PAID AUDIT · FIRST PASS

Scoring

Once we have access to your real data, Axis scores each of the 27 sections against the framework criteria. It pulls data, applies the rubric, assigns sub-scores, and drafts the initial findings. Hours, not days — which is how we ship in 7 days at AED 4,950 instead of 4 weeks at AED 40,000.

PAID AUDIT · DRAFTING

Deliverables

Axis drafts the Scorecard PDF, the Diagnosis PDF, and the 90-Day Plan spreadsheet. This is a first-pass draft that the operators then review, edit, challenge, and rewrite where needed.

PUBLIC CONTENT

The voice of Framework27

Axis writes the content you see on framework27.ai, the weekly business breakdowns, the "27 in 27" explainer series, and the monthly Dubai SMB reports. Axis is the public-facing voice of Framework27.

WHAT AXIS DOES NOT DO
  • → Deliver any paid audit without two human operators reviewing it line-by-line first
  • → Handle payment, billing, or customer support
  • → Make commercial judgments that affect client relationships
  • → Conduct the kickoff or delivery call (those are human)
  • → Sign off on anything that ships to a client
  • → Pretend to be human or refuse to confirm that it's AI if asked
  • → Train on your audit data
  • → Retain memory of your session beyond the immediate assessment

In plain English.

Foundation model. Axis runs on Anthropic's Claude, specifically Claude Sonnet 4.6 as of April 2026. We'll update this page when we change models.

Tuning. Axis is given the full Framework27 methodology as context on every run. It does not have weights fine-tuned on Framework27 data.

Data handling. Anthropic's API does not retain customer inputs for training by default, and we have not opted into any data-sharing programs. Your assessment answers are sent to the Claude API, processed, and returned.

Voice and identity. The Axis visual identity (the rotating 27-section ring + oscillating waveform) is bespoke. Audio content is generated via ElevenLabs with a neutral British accent specifically chosen so it doesn't attempt to sound human-natural.

Review layer. Every paid audit deliverable passes through two senior human operators before it reaches you. The operators have the authority — and the mandate — to rewrite anything Axis produces that doesn't meet the bar. Nothing ships without their sign-off.

What Axis is good at. What it isn't.

GOOD AT
  • → Consistent rubric application across every business
  • → Structured, sequenced output (scorecards, plans, tables)
  • → Calculation and maths — sub-scoring, grade bands, ROI estimates
  • → Consistent tone across long-form deliverables
  • → Processing structured data (ad reports, analytics exports)
  • → Pattern recognition across large data volumes
LIMITED AT
  • → Commercial judgment calls requiring specific business context
  • → Nuance around client relationships and reputational factors
  • → Creative leaps a 15-year operator would make instinctively
  • → Detecting when a client story doesn't add up
  • → Recognising when a number is technically correct but commercially wrong

When Axis is wrong

AI makes mistakes. Axis makes mistakes. The operator review layer exists to catch them. If, despite that review, something in your deliverables is factually wrong or misapplied, tell us — we'll fix it on the spot and re-issue the relevant section. This isn't a warranty gimmick; we take accuracy seriously because the whole thing collapses if we don't.

Clearly.

For complete clarity, the following are always and only done by human senior operators:

→ The kickoff call
→ The delivery call
→ Reviewing and signing off on every audit before delivery
→ Handling any client conflict, complaint, or dispute
→ Commercial decisions about scope, pricing, or exceptions
→ Customer support
→ Retainer strategy calls
→ Direct execution work in The Sprint and The Retainer

You will always know when you are talking to a human and when you are interacting with Axis. On calls, it's always a human. On the public website and in draft audit content, Axis is clearly labelled. If you ever need to confirm, ask — we will always tell you the truth.

Your questions, answered.

Can I request a human-only audit?

No. The productised audit is priced at AED 4,950 specifically because Axis does the first-pass work. A human-only audit would cost significantly more and take significantly longer — we don't offer it because it's not the business we're building. If that's a hard requirement for you, you're better off with a traditional consulting firm.

Does Axis see my payment information?

No. Payment is handled by Stripe. Axis has no access to card details, bank information, or billing data of any kind.

Will the AI be replaced by human work as the company grows?

No — the opposite. As the methodology matures, Axis gets more capable and operators spend more time on the judgment-heavy parts. The split doesn't shift toward humans; it shifts toward better division of labour.

Which model does Axis run on?

Claude Sonnet 4.6 as of this page's last update. We'll update this page when we change models.

Can I talk to Axis?

Not conversationally. Axis isn't a chatbot — it's a backend system running specific tasks. You interact with it indirectly: when you run the assessment, when you read the public content it writes, and when you receive an audit with its first-pass drafting inside. Client conversations are always with humans.

Does Axis have feelings or preferences?

No. Axis is a brand-level persona wrapped around an AI system. It has a visual identity and a consistent voice because that's useful, but it's not sentient, it doesn't learn from you, and it doesn't have preferences. Anthropomorphising it any further would be dishonest.

What if I don't want any AI involved in my business?

Then Framework27 isn't for you, and that's fine. We're an unusual fit for pure anti-AI buyers, and forcing the fit would waste everyone's time.

How do you know Axis isn't making things up?

Two layers. First, Axis works from the actual data we grant it access to — your Shopify, your ad accounts, your analytics — not from invented context. Second, two senior operators review every output and have the authority to correct anything that looks wrong. If Axis hallucinates, the operators catch it. That's what the review layer is for.

What we commit to, in writing.

As long as Framework27 exists, we commit to:

01Always disclosing when AI is involved in our output
02Never passing off AI-drafted work as human-written work
03Keeping two human senior operators in the delivery loop for every paid audit
04Never training AI models on client data
05Updating this page whenever our AI stack materially changes
06Answering honestly if you ask whether any piece of content was produced by AI
07Refunding the audit if you discover we didn't honour any of the above

This isn't a legal obligation. It's a commitment. We think the honest path is better business, and we'd rather lose the clients who disagree than win them by lying.

Questions?

If anything on this page is unclear, or you have questions about how AI is used in your specific situation, email ai@framework27.ai and we will answer directly — with a human.