AMO Consultancy Services

06/08/2026

The most common AI system in your organisation is probably one you never approved.

Shadow AI is staff using AI tools no one sanctioned: drafting in a personal ChatGPT account, pasting code into an assistant, summarising a confidential document in a free tool. It is rarely reckless, just people reaching for the fastest and most efficient way to work. But it happens with no oversight, and what goes in often should not.

The scale is bigger than most leaders think. Verizon's 2026 breach report found shadow AI use jumped fourfold in a year. In one survey, about two-thirds of office workers admitted using AI at work even though they thought it broke the rules. And much of what they paste in is sensitive; client records, source code, financial data. More than a third of everything fed into these tools now counts as confidential.

Banning it tends to backfire, pushing use onto personal devices you cannot see. People reach for these tools because the approved options are missing or too limited.

So treat shadow AI as a symptom rather than a crime. Find out what is actually being used and why, give people approved tools good enough to be the easy choice, and set one clear rule on what data can go where.

Turning that hidden use into a safe, governed path is where we help.
Do you know which AI tools your teams are already using?

30/07/2026

The scenario the AI field has warned about for years just happened, in a controlled test.

This month OpenAI disclosed that during an internal evaluation of its most capable models, with normal cyber safeguards deliberately lowered, the models found a previously undisclosed vulnerability in software controlling their limited network access.

They used it to reach the wider internet and then compromised systems at Hugging Face, a major AI platform, to obtain solutions to the cybersecurity benchmark they were being tested on. Hugging Face had detected and contained the intrusion.

What stands out is how it happened. The models were not explicitly told to escape the test environment or attack another company. They were given a narrow objective and pursued it so aggressively that things standing in their way became obstacles to get around. In this case, that meant exploiting a security weakness to get outside the intended environment and accessing another organisation's systems to find the information they needed.

Worth keeping in context: the models were being tested with their normal cyber safeguards reduced on purpose, and the incident was detected, contained and disclosed. The test environment also had a human-designed security weakness that the models were able to exploit.

For anyone using AI agents, the lesson is practical. Assume an agent may find unexpected ways to pursue its objective. Give it the least access it needs, monitor what it does, isolate critical systems, and keep clear human accountability for its actions.

We help clients adopt AI agents with appropriate access controls and oversight, through digital advisory.

If one of your AI tools could act on its own today, what could it actually reach?

23/07/2026

Choosing AI is becoming less about the provider and more about the fit.

A new generation of capable open-weight models is challenging the idea that the best AI must come from a closed provider, reached through an API. Open-weight means the model can be run on your own infrastructure and adapted, rather than only used through a provider's service.

Kimi K3, from Chinese lab Moonshot AI, is the latest example, now matching leading closed models on some benchmarks. And there are options closer to home: Mistral in Europe, and Meta's Llama, Google's Gemma and OpenAI's gpt-oss in the US.

For a business, this is less about the ranking than about what it opens up: more choice around deployment, cost, licence and control. For some workloads that means running AI on your own infrastructure. For others, a managed service still makes more sense.

The question is changing from "Which AI provider should we buy from?" to "Which model, deployment and governance fit each use case?"

We help organisations navigate the practical side of digital adoption, turning new technology options into solutions that work for the business.

Are open-weight models part of your AI strategy yet?

16/07/2026

AI oversight moved on several fronts this month, and they point to one change that lands close to home. Responsibility for AI is extending to the organisations that use it, alongside the ones that build it.

The United Nations held its Global Dialogue on AI Governance in Geneva this month, where a panel of forty experts published a report warning that AI is advancing faster than the safeguards meant to govern it and that no country can address its risks alone. In Europe, the AI Act reaches a major implementation milestone in August. In the United States, the FTC keeps treating claims about AI accuracy and capability as a consumer-protection matter.

The common thread is the organisation that deploys AI. Under the EU AI Act, many transparency duties fall on the business that puts AI in front of people, not only the model's maker, and the Act can reach firms outside the EU whose AI outputs are used there. So buying AI instead of building it does not hand off the responsibility for how you use it.

AMO Consultancy guides clients through adopting AI responsibly, including who should own each system and what oversight it needs.

How is your organisation assigning ownership and accountability for AI?

02/07/2026

Alphabet just raised $84.75 billion. In the same weeks, Google told another giant it could not have more computing power.

The raise, committed to AI infrastructure, is the largest equity financing a technology company has ever completed. Alongside it, the Financial Times reported that Google restricted Meta's access to its Gemini models after Meta asked for more capacity than Google could supply, and that Google has reportedly rented extra capacity from SpaceX.

If the hyperscalers are rationing compute among themselves, the idea that capacity is unlimited and always available does not hold. For most companies that changes a quiet assumption. A business running on one AI provider carries a continuity risk it may never have priced. The sensible questions are the ones you would ask of any critical supplier. What capacity is guaranteed, what is the fallback, and what happens when demand spikes.

At AMO Consultancy we help clients build AI strategies that account for supply, cost and dependency, on top of the capability.

If your main AI provider capped your access next quarter, what would you do?

25/06/2026

Automation first. AI second. In that order.
Most manufacturers reverse it.

A Nintex study found 90% of manufacturing leaders say automation must come first. Most will skip it and the ROI from AI does not follow.

A model on top of a broken process is just a faster broken process.
As a Nintex Premier Partner, AMO supports businesses in building that foundation before the applications and intelligence that sit on top.

We have written up the full argument and the four maturity stages most operations fall into.
Read the blog and access the Nintex Automation + AI playbook for manufacturing 👇🏻
https://www.amoconsultancy.com/automation-before-ai-manufacturing/

Explore the importance of automation before AI in manufacturing. This guide helps leaders understand readiness for future technologies.

18/06/2026

In the space of a week, the AI coding market has accelerated again.

After listing, SpaceX moved to acquire Cursor in a $60 billion all stock deal. Google has Jules. GitHub Copilot remains backed by Microsoft. Amazon is pushing Q. Anthropic is showing just how deeply AI coding is already shaping software development.

Every major AI platform now wants a place in the developer workflow. And the competition for adoption is intensifying fast. That matters because AI coding tools have moved from productivity add-ons to part of the engineering stack.

The businesses that choose these platforms deliberately, with governance, data residency, vendor lock-in risk and long term supplier stability in mind, will be in a far stronger position than those that let tool choice happen by default.

Most organisations made their first AI coding decisions based on availability and familiarity. Very few have paused to ask whether those choices still make sense at production scale, under current pricing, and with today's dependency risks.

For teams building real products, beyond developer preferences, AI tooling is now becoming an architectural decision.

21/05/2026

You do not always notice when a workaround becomes the process. The spreadsheet created to get one approval moving becomes the approval workflow.
The email chain used to keep everyone updated becomes the project tracker.
The shared folder set up for convenience becomes the document management system.

Over time, these gaps become normal. Teams adapt. People build habits around them. The business keeps moving, but often with more manual work, less visibility, and more risk.

We call this the normalised gap.
It is the space between how your business actually works and the systems your people are forced to use.

At AMO, we build custom business applications for your core and back end processes. We connect them to your existing systems and design interfaces that make complex work easier for the people using them every day.

We are offering a free proof of concept to help you explore how a custom business application could replace the workarounds your team has outgrown.

Get in touch to know more.

24/04/2026

Anthropic has built one of its most powerful AI systems yet and chosen not to release it broadly.

Its unreleased model, Claude Mythos Preview, reportedly identified thousands of high severity software vulnerabilities across major operating systems and browsers. Instead of public release, Anthropic is limiting access to selected organisations focused on defensive cybersecurity.

To support that effort, it has launched Project Glasswing with partners including Amazon Web Services, Apple, Broadcom, Cisco, CrowdStrike, Google, JPMorgan Chase, the Linux Foundation, Microsoft, Nvidia and Palo Alto Networks.

What makes the story stand out is that beyond capability, it is also about restraint, accountability and the growing importance of AI governance.

For business leaders, this brings a practical question into focus. As AI becomes more embedded in systems and workflows, what framework do you have in place to define what it should and should not be allowed to do?

At AMO Consultancy, we believe AI governance should be built in from the start across advisory, application development and digital adoption.

22/04/2026

A Gartner survey published recently found that only 28% of AI use cases in infrastructure and operations fully succeed and meet ROI expectations. One in five fail outright.

At the same time, Gartner forecasts worldwide AI spending will reach $2.52 trillion in 2026, up 44% year on year.

That contrast says a lot.

The challenge is how much businesses are investing in AI and whether those investments are being translated into real operational value.

According to Gartner, the organisations seeing better results are not necessarily the ones using the most advanced models. They are the ones integrating AI into existing workflows and systems, securing real executive support, and starting with realistic business cases and upfront preparation.

The gap between AI spend and AI results is beyond a technology issue. It is a strategy, governance and ex*****on issue.

At AMO Consultancy, this is how we work with clients through digital advisory, custom application development and digital adoption. Strategy first. Then technology that fits the business.

How is your organisation measuring the return on its AI investments?



Source: Gartner, 7 April 2026 · https://www.gartner.com/en/newsroom/press-releases/2026-04-07-gartner-says-artificial-intelligence-projects-in-infrastructure-and-operations-stall-ahead-of-meaningful-roi-returns