Agentic Design Team

An organisational design in which AI agents perform a meaningful share of the work, and the structure, roles, decision rights and governance are rebuilt around that fact rather than bolted onto the existing hierarchy. It specifies what humans own, what agents own, how agent decisions are audited, and who is accountable when an agent gets something wrong. AI-native organisation design is another accurate name for the same work.

Yes. The deliverable is a target operating model in the same sense a Big Four engagement uses the phrase: structure, roles, decision rights, governance and the sequenced plan to move to them. Two things distinguish ours. It is designed for an organisation in which AI agents perform a meaningful share of the work, and the technical architecture is designed alongside it by the same pair, because a target model the platform cannot support is a document rather than a design. If you have bought TOM work before, the shape will be familiar; the decision rights will not be.

Because a target operating model the platform cannot support is a document rather than a design, and an architecture built without knowing which decisions move to agents optimises for the wrong things. Separating the two is one of the most reliable ways to produce a programme that stalls at the point of scaling.

Yes, with one caveat about scope. Designing how the organisation works around agents is one job, and running the delivery is another, though most organisations advertise the two together as Head of AI. This rung supplies the first, as an operating-model lead alongside an agentic architect, for two to four months. If what you need is the delivery half, that is programme and delivery management at £18,000 to £35,000 a month, or the Agentic Build Team if the thing also has to be built. Either way it is an engagement rather than an appointment, and writing the specification you recruit against is part of the work.

The two disciplines are different. Operating-model design is about decision rights, accountability and adoption; agentic architecture is about platform, data, integration and security. One person covering both does one of them badly. Two senior practitioners is the smallest team for the work.

By the consequence and reversibility of the decision, not by task complexity. Agents take decisions that are high-volume, observable and cheaply reversible. Humans retain decisions that are consequential, contested, or hard to undo. The boundary is written down explicitly per role, and the escalation path across it is part of the governance framework.

Design only, and deliberately so. This rung designs what gets built rather than building it, so there is no build engineer on it. The two seats are an operating-model lead and an agentic architect, and the architect gets the architecture signed off with your CTO rather than handing it to them. What you get is a target operating model with roles defined by the decisions they own, the platform, data, integration and security architecture, a governance framework that makes agent decisions auditable rather than theoretical, human-in-the-loop boundaries defined per decision class with escalation paths, and a sequenced build plan. Building it is a different rung with a different team.

From the published rate card, not from a negotiation. The team is two senior practitioners, an operating-model lead and an agentic architect, both at the published senior rate of £1,250 a day, and engagement prices derive from the rate card at twenty billable days a month, excluding VAT. Two senior seats on that basis sit inside the published band, and where a given month lands in it follows how much of each seat the work needs. There is no junior pyramid underneath the pair. It bills monthly, typically over two to four months, and either side can end it on 30 days' written notice.

The work runs in four phases. Weeks one to three establish current-state truth: how decisions actually get made, what your data and platform will support, and where the structure will fight the technology. Weeks four to eight design the operating model and the technical architecture together. Weeks nine to twelve cover governance and assurance: audit trails, escalation paths, model risk, and the human-in-the-loop points your risk function and your regulator will both ask about. The final weeks are sequencing and handover, with the build plan, the adoption plan and named internal ownership for every element. A shorter engagement runs the same sequence compressed, and every month is expected to show measurable value.

No. The engagement ends on a sequenced build plan your own teams, ours, or a third party could execute, the architecture is signed off with your CTO rather than handed to them, and you own all the work product. If you want Tenhaw to execute it, the next step on the ladder is the Agentic Build Team, three people at £70,000 to £85,000 a month; if you would rather build with the engineers you already have, the plan is written to be executable without us. The design work is priced to stand on its own, not to seed the next sale.

That is one of the clearest cases for this engagement. When several functions build independently you get incompatible platforms, governance that contradicts itself and integration work paid for twice, and unpicking that afterwards costs more than designing it once. The pair, an operating-model lead and an agentic architect, set the platform, data foundations, integration and security architecture centrally, define which decisions move to agents and who is accountable for them, and end on a sequenced build plan your functions execute against rather than around. It runs at £35,000 to £55,000 a month, typically over two to four months, and the architecture is signed off with your CTO rather than handed to them.

Probably, and we would rather say so now. This rung is organisation-level design. It earns its price when several functions are affected, when structure and platform decisions have to be taken together, and when governance has to exist before anything scales. A single-team pilot does not need a target operating model, and neither does an organisation that has not yet established where agents create value. Better starting points are the AI Readiness Audit, four weeks at £44,000 fixed, which finds where AI does and does not pay, or an Agentic Proof of Concept at £20,000 to £55,000 over two to four weeks. Come back to design work when scaling is the actual problem.

By making each month end in something you would still want if we stopped there, rather than in progress towards a document at the end. The test is whether a decision your leadership was stuck on can now actually be taken: which decisions move to agents, what the platform has to support, what your risk function will accept. Design work goes wrong when all of its value arrives in the final week. Engagements here are retainer-shaped rather than milestone-shaped, and a month that delivers no measurable value is reported as a failed month rather than folded quietly into the next one.

Then it is not finished, and that is deliberate. The architect signs the architecture off with your CTO rather than handing it to them, so disagreement is meant to surface during weeks four to eight, while the operating model and the architecture are still being designed together, rather than at a final readout. An architecture your CTO will not own is a document rather than a design, and it will not survive the first real build decision. If the disagreement turns out to be irreconcilable, the engagement is monthly and cancellable on 30 days' written notice either way.

Between roughly £70,000 and £220,000 excluding VAT, depending on how long it runs. The pair is priced at £35,000 to £55,000 a month and the work typically takes two to four months, so two months at the bottom of the band is about £70,000 and four months at the top about £220,000. Where a given month lands in the band follows how much of each seat the work needs, and there is no junior pyramid underneath the two. It bills monthly rather than as a lump sum, either side can end it on 30 days' written notice, and nothing obliges you to build with us afterwards.

Everything designed to that point, and all of it is yours. The engagement is retainer-shaped rather than milestone-shaped, so each month is expected to deliver measurable value on its own, and a month that delivers none is reported as a failed month rather than folded into a later deliverable. After the first weeks you would hold the current-state work: how decisions actually get made across the organisation, what your data and platform will genuinely support, and where the structure will fight the technology. That is useful whoever does the rest of it. Notice is 30 days either way, in writing, and you own all the work product.

If they also hold a mandate over how the organisation works, yes, and you should keep the money. Capability is rarely the constraint. Most enterprise architecture functions own platform, data, integration and security and carry no authority over decision rights or accountability, so the architecture gets designed against an operating model nobody has agreed, which is why this rung is a pair rather than one architect. Where your architects are strong, the better shape is designing alongside them rather than around them, because they know the estate better than we will in three months and they are still there after we go.

Often, and it is the first thing we test. Weeks one to three set current-state truth against how decisions actually get made, so a model that is honest about structure and accountability is a real head start. Two things usually have to be rebuilt, because the model was never written to answer them. Roles are defined by the tasks people do rather than by the decisions they own, and nothing in it draws a boundary between what a human decides and what an agent decides. Where the existing model is sound the design phase is narrower and the engagement lands nearer two months than four. Bring it to the first conversation, which is the cheapest way to find out how much survives.

Concentrated in weeks one to three, when Tenhaw's operating-model lead establishes how decisions actually get made across the functions in scope, and that is leadership's own time rather than something the pair can do on their behalf. Your CTO carries the heaviest share, because the architecture is signed off with them across weeks four to eight rather than presented at the end, and risk and audit arrive around weeks nine to twelve when the governance framework is written. The engagement runs two to four months at £35,000 to £55,000 a month, and the larger demand is not diary time. The design forces decisions leadership has been deferring. How many hours that costs follows how many people hold those decisions today.

That is one of the things weeks one to three exist to surface. Tenhaw's operating-model lead maps how decisions actually get made today rather than how the organisation chart says they are made, so a contested one shows up in the current-state work instead of at a final readout. Accountability is then mapped across the model before a single agent is deployed, and the final weeks attach named internal ownership to every element, so the decision leaves with one person against it. Where it stays genuinely contested, the pair cannot grant authority your executive committee has not settled, and that is the part a two to four month engagement cannot compress.

Beside it, feeding it. Tenhaw's design pair carries no build engineer and no delivery management, so instead of competing to run your programme, it produces the target operating model, the architecture your CTO signs off, the governance framework and a sequenced build plan your programme executes against. The place the two collide is the plan, with two roadmaps pulling at the same teams, which is why the engagement ends on one sequence with named internal ownership for every element. If the programme also needs someone running delivery, that is a separate rung at £18,000 to £35,000 a month. Whether the AI work sits inside your programme governance or alongside it turns on who holds the platform decisions.

Yes, provided the division genuinely holds its own decisions and its own platform. Tenhaw prices it at £35,000 to £55,000 a month whether the scope is a division or a group, because this is organisation-level design rather than team-level, and a division that owns its data, its integration points and its budget meets that test. It goes wrong where a division has to design on top of group platform, identity and security decisions it does not control, because the model then encodes choices someone else can overturn. Weeks one to three apply that test directly, and whether it passes turns on how much of the platform your group holds centrally.

Most procurement teams want a ceiling and an end date, and both exist here. Tenhaw's design pair runs at £35,000 to £55,000 a month across two to four months, so a purchase order can be raised against the months you have approved rather than as an open commitment, and the exit date is agreed at kickoff rather than left to drift. The contract is monthly and cancellable on 30 days' written notice either way, and you own all code, documents and work product as it is produced. What your finance team treats as committed spend usually turns on whether they book the notice period alongside the approved months.

Scale sits in the sequencing rather than in how many people draw the model. The operating-model half of Tenhaw's pair has run at that scale, on a target operating model co-designed, piloted and proved for 500 teams at HSBC, in James Rooney's transformation role there, with global rollout due in 2026 and not yet rolled out. The pair designs and your own people carry the volume, which is why weeks one to three test how decisions actually get made across the functions in scope, and the final weeks attach named internal ownership to every element. What sets the workload is the breadth of scope rather than the size of the organisation.

Rarely, because this rung designs rather than builds. Tenhaw's agentic architect owns platform, data foundations, integration and security, and that design work runs off documentation rather than live systems: architecture and integration documentation, data catalogues, identity and security policies, and read access to schemas rather than to production records. Where something genuinely has to be seen live, it is seen on your own infrastructure under your own policies, with UK data residency by default and EU available, and the 24-hour personal-data-breach notification sits in the DPA either way. Both practitioners are screened to BS7858 before any client access. How much of the estate opens at all follows how far the architecture reaches.

Usually yes, and often they should, because two to four months with nothing shipping is its own risk. Tenhaw's pair has no build engineer in it, so nothing on this rung stops your teams delivering. The care is needed on irreversible commitments: platform selection, identity and access patterns, the integration layer, anything expensive to unpick if the architecture lands somewhere else. Those are what the agentic architect settles with your CTO across weeks four to eight, so they are worth holding. Reversible work carries on as normal. Which of your in-flight items sits on which side of that line is one of the things weeks one to three sort out.

It becomes a constraint the architecture is designed around, and weeks one to three test how hard a constraint it is. Tenhaw's agentic architect owns the platform, data foundations, integration and security half of the pair, and works from what your estate will genuinely support, so a platform you have already bought is taken as given unless it cannot carry the decisions the operating model moves to agents. Where it cannot, that gap becomes sequenced work in the build plan rather than an assumption nobody checks. Your CTO signs the architecture off, so any decision to replace something is theirs. Which way it lands depends on what the platform is being asked to do.