From AI translators to AI operators

In a recent article on Unhyped AI, Stuart Winter-Tear, argues that AI not only needs people who can explain the technology in compelling language but it also needs people who can turn capability into operating decisions. This debate about what is termed “AI Translation Debt” captures what many senior organisational leaders feel but rarely name - the unpriced work of stitching together silos, clarifying rules and reconciling conflicting metrics that sits between policy, IT and frontline reality. Many organisations now accept that their models can perform technically; what is missing is the organisational muscle to let them operate with real autonomy, without piling on extra layers of checking and governance.

The issue is no longer a translation problem between “the business” and “the tech”; it is a problem of operating discipline. Leaders need people who can move beyond slideware into hard service levels, shifting from lists of “AI use cases” to redesigning how decisions, risk and accountability actually move through organisations.

Leadership is still the bottleneck

In a recent piece titled “AI won’t fix timid leadership” we argued that most AI programmes stall not because the technology fails, but because leadership behaviours do not change. They insist on “safe experiments” that live in innovation teams and sandboxes, never touching real services, budgets or risk.

The same pattern shows up in the translation debate. Organisations try to hire their way out of AI Translation Debt with a new role, job title or centre of excellence, instead of confronting the real question: who is prepared to re-make structures, incentives and hand‑offs so AI can actually carry work, not just generate drafts?

A case study in AI Translation

A good example of what AI Translation looks like in practice is Hey Geraldine, developed by Datnexa in partnership with Peterborough City Council. The tool was created to address a very specific operational problem in adult social care: valuable frontline expertise was concentrated in one highly experienced occupational therapist, while wider teams still needed fast, reliable guidance on technology-enabled care and best practice.

This is exactly the kind of “AI translation” challenge too many councils misunderstand. The issue was not a lack of information, nor a shortage of interest in AI. It was that expert judgement existed, but it was trapped in people, routines and informal workarounds that could not scale across shifts, teams or rising demand.

Datnexa’s role was not simply to build a chatbot. Datnexa helped turn that expertise “from page to product” by creating a secure, governed knowledge base grounded in local policy, practice and practitioner insight, then supporting the council through DPIAs, data processing agreements, service agreements and wider information governance processes before testing began. In other words, the work of translation was operational as much as technical: making expert knowledge usable, governable and trusted in the flow of real frontline work.

The result was a 24/7 AI assistant that gave practitioners instant access to guidance in the tools they already used, including Microsoft Teams, rather than forcing them into another disconnected system. For organisational leaders, AI translation is not about placing a clever interface on top of existing confusion; it is about capturing scarce expertise, aligning it to local governance, and embedding it in a service model that staff can actually use. 

This is where Datnexa is different: not in offering generic AI capability, but in turning hard-won frontline expertise into governed, usable systems that organisations can deploy safely and scale with confidence.

What effective AI translators really do

If you strip away the hype, the most valuable “AI translators” inside a council do three hard things that rarely fit into a job description.

• They name the hidden work: spelling out where meaning gets lost, where policies contradict real practice, and where managers quietly perform reconciliation work that never appears on an org chart.

• They reduce hand‑offs: collapsing fragmented processes across directorates so that machine speed is not neutralised by human queueing, checking and re‑work.

• They reset metrics: shifting attention from model accuracy in a lab to lived outcomes such as fewer hand‑offs, shorter backlogs, less after‑hours firefighting and clearer accountability.

That is not “translation” as in telling engineers what the service wants. It is operational design. It demands permission to touch span of control, budgets, policies and line management, not just prompts and process maps.

How Datnexa does this with councils

Datnexa was created to work in this uncomfortable space between AI capability and organisational readiness, with a particular focus on local government and public service delivery. We do not sell platforms, and we do not treat “translators” as a thin layer between two unchanging worlds. Instead, we put small, multidisciplinary teams alongside your directors, heads of service and service owners to design for humans first, then automation.

A typical Datnexa engagement with a council focuses on four moves:

• Expose the translation debtWe map where work actually flows today - across contact centres, line of business systems and partner organisations - not how it appears on a process diagram. This makes the cost of doing nothing visible in the language of missed KPIs, budget pressure and staff exhaustion.

• Redesign a real service, not a labRather than another proof‑of‑concept in a sandbox, we pick a concrete service like housing repairs, revenues and benefits, or adult social care front door, and redesign the work around AI from end to end. The goal is fewer steps and clearer accountability, with AI taking on the stitching work that currently drags down your best people.

• Re‑shape leadership behavioursWe work with corporate leadership teams on decision rights, escalation pathways and risk appetite, so that autonomy does not simply create new checking loops. This is where timid leadership shows up fastest—and where our job is to help leaders make a small number of explicit, testable commitments instead of broad, cost‑free endorsements of “innovation”.

• Instrument for lived outcomesFinally, we help you track changes your residents and staff can actually feel: fewer hand‑offs between teams, shorter backlogs, fewer repeat contacts, and decision‑making that matches reality instead of templates. These become the success measures for AI, not just model performance curves or licence utilisation.

When to call Datnexa

If your organisation recognises itself in the idea of AI Translation Debt - multiple disconnected pilots, growing governance overhead, very little visible change in how services actually feel - then your challenge is not another model, it is operating translation. That is where Datnexa is strongest.

We work best with senior leaders who are prepared to treat AI as a chance to redesign work, not just redecorate it. If that is you, the smartest move is to start with one high‑stakes, high‑friction service - like housing repairs, adult social care front door, or revenues and benefits - and commit to redesign it together.