ADQ has named four value-creation drivers — sustainable investing, talent, digital transformation, and innovation and R&D. Each is well-defined and well-resourced. The harder part, at a portfolio of 25+ companies and eight clusters, is the day-to-day connective work that lets those drivers compound across the portfolio rather than play out in isolation at each company.
We can help with that connective work using Agentic AI — small, focused Agents that read from the systems each portfolio company already runs, share what should be shared under written policy, and keep an evidence trail intact by construction. This paper sets out where we believe we can help, what it would look like for the people on the receiving end, and a 120-day engagement at AD Ports Group that proves the value before any portfolio-wide commitment is made.
The engagement is small on purpose. We would rather earn the right to do more.
ADQ has been clear about what creates value across its portfolio. What is harder is making the four drivers work *together* — so that a digital initiative at one portfolio company strengthens the innovation pipeline at another, so that talent insight in the Strategy Office reaches the place it can be acted on, so that ESG commitments hold up under audit because the evidence sits where the decisions were made.
This paper describes how we think we can help, where to start, and what we would ask in return. It is deliberately small at the front end, with a clear path to portfolio-wide scale if it works.
ADQ has named four value-creation drivers. The next section ("What this looks like in practice") gives each one a face — a person on a Tuesday morning whose work is shaped by it. Each driver below points to whose morning shows the work in practice.
*This is Sara's morning, and Hala's morning. Sara works inside one portfolio company; Hala extends the same working pattern across the cluster boundary to a second.*
ADQ has named the goal: a portfolio that is modern, high-performing, and future-proof. The harder version of that goal is making digital transformation compound across the portfolio, so that each acquisition or new venture starts from where the last one ended rather than from zero.
We can offer a shared way for portfolio companies to make their operational signal visible to one another and to the Strategy Office, without replacing any of the systems they already run. AD Ports, Aramex, Etihad Rail, and Etihad Airways being able to see each other's operating tempo in real time — so a container hand-off or a freight booking becomes a piece of information the next portfolio company can use. New acquisitions joining that picture from day one. The Strategy Office seeing the operating health of the portfolio in days, not quarters.
We are not proposing to rebuild any portfolio company's stack. We are proposing to add Agentic AI that respects each company's data boundaries and gives ADQ a portfolio-level view that does not exist today.
*This is Layla's morning. A long-term thesis that comes alive with continuous evidence rather than waiting for the quarterly memo.*
ADQ has been deliberate about taking a long-term view. The challenge with long-term commitments — the $25B Energy Capital Partners partnership, the critical-minerals work with Eni, the seed development with Limagrain, the infrastructure work through Gridora and Plenary — is that each rests on a thesis that needs to keep being true over years. The cost of a thesis quietly decaying is far higher than the cost of a thesis being wrong from the start, because the capital has already been deployed.
We can help the Strategy Office track each of those commitments as a living thing — continuously fed with signal from inside the portfolio and outside it, by Agents whose job is to score the thesis as the evidence changes. The goal is not to replace the strategy team's judgement. It is to make sure the team is the first to know when a thesis is changing, not the last. Over time, this also makes ADQ's innovation pipeline more legible to its board, its partners, and its co-investors.
*This is Mariam's morning. A regulator's question that used to consume a quarter, answered before lunch.*
ESG commitments are easy to declare and hard to defend. As ADQ deploys capital into international markets — Vietnam, Azerbaijan, Central Asia, the United States — the bar for demonstrable, auditable evidence behind sustainability claims keeps rising. Investors, regulators, and partners increasingly want claims they can verify, not narratives they have to trust.
The work we do can help here in a quiet but valuable way: every operating decision and every relevant data point that flows through the Agentic AI carries its own evidence with it — when it happened, where it came from, who acted on it. That means an ESG report, a regulatory disclosure, or a board update is a query, not a research project. Over time, it gives ADQ something genuinely scarce: a sustainability story that holds up to inspection across every portfolio company and every jurisdiction it operates in.
*This is the through-line in all four mornings. Every Agent that works alongside Sara, Hala, Layla, and Mariam learns from how each of them responds — so the case-pattern memory at the port, the thesis history at the Strategy Office, and the compliance evidence at the audit interface all grow with use. The institutional knowledge that used to leave with rotating staff stays with the firm.*
Across 25+ portfolio companies, the practical question is how to keep institutional knowledge from leaving with the people who hold it. A senior portfolio manager who has worked through three deal cycles knows things a new hire cannot read in a deck. A compliance investigator who has closed hundreds of cases sees patterns no policy document captures. A thesis manager has spent two years refining the assumptions behind a $25B commitment.
We can help capture that working knowledge — not as documents in a folder, but as context that surfaces at the moment a decision is being made. Over time, this becomes part of ADQ's competitive position: the firm gets better with each cycle, instead of starting over with each new generation of staff.
The four drivers above describe what we believe we can help with. The four journeys below describe what it actually feels like for the people on the receiving end. Four people, in four roles, on the same Tuesday morning. The same Agentic AI underneath all of them.
*Driver 1 — Digital transformation, inside a portfolio company.*
It is 7:42 on a Tuesday at Khalifa Port. Sara is the operations manager on duty.
Today, without the Agents. Sara has six screens open — terminal operating system, gate system, customs interface, vessel schedules, weather, email. The crew radio crackles. A driver is waiting at Gate 4 for a container that has not cleared customs. Berth 3 has been turning slow for two days but no one has yet noticed the pattern. By 9:00 she is on the phone explaining a slipped schedule to a shipping line.
With the Agentic AI. Sara opens one screen. The Agents have been working all night. A handful of things are waiting for her attention, ranked.
A signal-handling Agent has noticed that Berth 3's dwell time has been creeping up across five vessels — not enough for any single alarm to fire, but enough to be a pattern. It shows Sara the pattern, the vessels involved, the customs windows in play, and adds a plain-language note: *"Three similar patterns last quarter were resolved by re-sequencing the gate roster. Want to see those cases?"* Sara looks. She decides. She acts.
A second Agent, watching customs hand-offs, has already flagged the Gate 4 container thirty minutes ago and pinged the customs team. Sara knows about it before the driver does.
Sara still runs the port. She still makes every call. The Agents just make sure the right things reach her at the right moment — and the longer she works with them, the more they learn what *she* considers a problem.
What changed. She used to spend her morning catching up. Now she spends it ahead.
*Driver 1 again — Digital transformation, this time across the cluster boundary.*
Hala runs middle-mile dispatch at Aramex. She sends drivers to the port to pick up containers.
Today, without the Agents. Half the time the container is not ready — not cleared, not discharged, not moved to staging. Drivers wait. Trucks idle. Hala builds slack into every pickup window because she has to. The slack costs her hours and her drivers' patience. She has a planned pickup time, and a different actual reality, and no real visibility into the gap.
With the Agentic AI. A perishable container is on a vessel discharging at AD Ports at 10:20. AD Ports' Agents already know — one has been tracking it. That Agent publishes a single signal into the cluster channel: *container X will be at staging zone Y around 11:05.*
What crosses the boundary is controlled, written down, and enforced. The container identifier and its availability window cross. The shipper's identity, the commercial terms, and any customer information do not. Aramex sees what Aramex needs. Nothing else.
Hala's dispatch Agents subscribe to that channel. Her routing Agent updates the pickup window on its own. The driver who would have left the depot at 9:30 now leaves at 10:00. He arrives at 11:08. The container is there. He is back on the road in twenty minutes.
Multiply that by every container, every day, across the Transport & Logistics cluster.
What changed. The driver still drives. Hala still dispatches. But two portfolio companies are now coordinating as one operation, without anyone picking up the phone — and ADQ's Strategy Office sees the coordination happening, in real numbers, in real time.
*Driver 2 — Innovation and R&D, keeping the long-term view honest.*
Layla is a thesis manager in the ADQ Strategy Office. She watches the Energy Capital Partners partnership — the $25 billion bet on US power generation feeding data-centre capacity feeding AI compute demand feeding critical-minerals demand feeding ADQ's broader positioning. It is a five-link thesis chain over a multi-year horizon. Her job is to know whether the thesis is still true.
Today, without the Agents. Quarterly memos. Industry reports she reads on flights. Calls with the ECP team. An analyst on her staff who tracks the upstream end. Another who tracks the downstream. By the time the chain reassembles in her head, three months have passed. By the time the next memo reaches the chairman, six months have. Decay happens in the gaps. More than once she has read in the Financial Times what she should have known a quarter earlier.
With the Agentic AI. Layla opens her interface on Tuesday morning. The thesis is on screen as a living chain — five links, each one with the latest evidence and a plain-language statement of what is currently true.
An external-signal Agent has been reading public filings, regulatory dockets, industry press, and supply-chain trackers overnight. It picked up a delivery-schedule shift on gas-turbine orders from a major US OEM — eighteen months later than the previous quarter's expectation. The downstream link of the thesis has been re-scored automatically: data-centre build-out across two key states slips on the new timeline. The link below that — AI compute demand — has not yet been re-scored, because the Agent is not certain the demand side will absorb the slip.
The interface shows her the evidence. The OEM's filing. The two analyst notes that confirm it. The regulators who signed off. The downstream impact. And the open question: does AI compute demand shift with it.
She has seen this before lunch on Tuesday. Not after the quarter closes.
She decides what to do — call the ECP team, escalate to Strategy Office leadership, or ask the Agent to track a specific question on the AI demand side. The Agent records what she asked and why, so the next person who picks up this thesis sees not just the evidence but the chain of judgement.
What changed. The thesis used to live in a quarterly memo. Now it lives in a continuously-updated picture that the Strategy Office can act on in days. And the chain of Layla's judgement becomes part of ADQ's institutional memory — there for the next thesis manager who picks up this work.
*Driver 3 — Sustainable investing, the evidence trail by construction.*
Mariam is a compliance officer. On Tuesday morning, an inquiry arrives: *"For the last quarter, show every operational decision that touched a sanctioned-jurisdiction route, the basis for approval, and the governance trail."*
Today, without the Agents. This is a six-week project. Her team digs through inboxes, terminal logs, customs records, and change-management tickets. Half the evidence does not exist in a queryable form. The other half exists but cannot be tied together. They write a memo explaining what they think happened, supported by what they can prove. The regulator accepts it because they have to. Mariam is not proud of it.
With the Agentic AI. Mariam opens the evidence interface and types the question close to her own words. An evidence Agent translates it into a structured query against the captured record of every event, every decision, every cross-boundary access, and every Agent action over the period.
By 11:15 she has the answer. Every relevant operational event, timestamped and traced to its source. Each one linked to the policy under which it was processed. Agent decisions tied to the model versions and the input context that produced them. Cross-boundary enforcements logged with reason codes. Nothing has been retrofitted. The records were created at the moment each decision was made.
She exports the report in the format the regulator expects and sends it before lunch.
What changed. The regulator's question used to consume Mariam's quarter. Now it consumes her morning. And the evidence is stronger — because it was never assembled. It was always there.
*Driver 4 — Talent and institutional knowledge, working as the through-line.*
Sara, Hala, Layla, and Mariam never speak to each other. They work in different roles, different portfolio companies, different time horizons. Sara runs operations. Hala runs dispatch. Layla watches a multi-year thesis. Mariam answers regulators.
But the Agentic AI underneath all four is the same connective layer — reading from the systems each of them already runs, sharing what should be shared under written policy, keeping the evidence trail intact by construction, and learning from how each of them responds. Sara's case-pattern memory at the port, Hala's coordination history with Aramex, Layla's chain of thesis judgement at the Strategy Office, and Mariam's library of compliance precedent are the same kind of asset, accumulating in the same kind of way. Institutional knowledge that today walks out the door with rotating staff stays with the firm.
For ADQ, this is what the four value-creation drivers look like in practice — not as a transformation programme that replaces what works, but as a thin connective layer that makes what already works work together, and makes the firm itself a little smarter every day it runs.
We do not think the right first step is a portfolio-wide programme. The right first step is a single, well-scoped engagement that lets us prove, in 120 days, whether what we have described is real and useful. We recommend starting with AD Ports Group, because it already operates a significant digital footprint and it sits at the centre of the Transport & Logistics cluster's signal flow.
That Agentic AI can do the connective work behind ADQ's value-creation vision — alongside the people already running the business, without replacing the systems they already use, with measures of success agreed up front, and with a clear path to scale to the rest of the portfolio if it works.
The three deliverables are not parallel workstreams. They are one integrated system, where each piece earns the next: the operational Agents at AD Ports (Deliverable 1) produce the signal that the cross-portfolio Agents share with Aramex or Etihad Rail (Deliverable 2), and the evidence trail (Deliverable 3) is the audit-grade record of everything Deliverables 1 and 2 do — created at the moment each event happens, not assembled after the fact. Each deliverable maps to one of the user journeys above.
#### Deliverable 1 — Agentic AI working alongside the AD Ports operations team
*The working version of Sara's morning.*
A small set of Agents read from AD Ports' existing operational systems and surface patterns, anomalies, and recommendations to the operations team in real time:
What we build, technically: read-only connections from AD Ports' existing systems (using our Scouts component); typed events for the chosen operational area registered in our LogOS event log so every observation has a schema, a source, and a timestamp; a boundary policy enforced by our PACT component defining exactly what the Agents can read; and a simple, read-only operator interface for the team.
What AD Ports has at the end: a live view of operational tempo across one chosen area, ranked operator-actionable insights, and a growing case-pattern memory the Agents draw on.
What this deliverable produces for the next two: a stream of typed, sourced, time-stamped operational events. Deliverable 2 reaches into that stream to share what should be shared. Deliverable 3 reaches into that stream to remember what happened.
#### Deliverable 2 — Cross-portfolio Agentic AI exchange between AD Ports and Aramex (or Etihad Rail)
*The working version of Hala's morning. Built directly on Deliverable 1.*
The same Agents that surface signal *inside* AD Ports (Deliverable 1) now share a controlled subset of that signal *across* the cluster boundary to a second portfolio company. With the longer 120-day window, we commit to two cross-portfolio signal types rather than one, so the pattern is demonstrated on more than a single example.
The Agents we add:
What we build, technically: a durable, typed signal channel between the two portfolio companies (LogOS outbox pattern) so each signal is committed to AD Ports' own log before it is delivered, and is exactly-once and immutable; boundary policies in PACT defining what crosses, what does not, and the reason-coded enforcement record; and a real-time view, for the ADQ Strategy Office, of how many cluster-level exchanges happen and what their indicative operational value is.
Why Deliverable 2 only works because of Deliverable 3. Cross-portfolio signal exchange is only acceptable to two portfolio companies (and to ADQ) if every exchange is boundary-controlled, policy-enforced, and auditable. The provenance and policy records described in Deliverable 3 are not separate from this exchange — they are what makes the exchange trustworthy in the first place.
#### Deliverable 3 — Audit-ready evidence trail for everything Deliverables 1 and 2 do
*The working version of Mariam's morning. Built by construction from Deliverables 1 and 2.*
Deliverable 3 is not a separate workstream. It is the byproduct of doing Deliverables 1 and 2 correctly: every event the Agents observe, every recommendation they make, every signal that crosses the cluster boundary, and every policy decision that allows or blocks an access — all captured in a queryable record, at the moment it happens.
What we add specifically for Mariam's experience:
What is already there because of Deliverables 1 and 2: every operational event from Deliverable 1 carries schema, source, timestamp, payload digest, and the chain of prior events that caused it (LogOS + outbox pattern); every Agent decision is bound to the model version that produced it, the input context, and the reasoning chain (Cognition Gateway); every cross-boundary access from Deliverable 2 is logged with a policy identifier, an enforcement decision, and a reason code (PACT).
What we wrap around it for the auditor experience: an export interface in a format external auditors accept on first review, plus a sample question demonstration where a query like *"show me every cross-portfolio signal exchange in the last 60 days, with full provenance and policy enforcement"* is answered in seconds.
Why this matters for ADQ at portfolio scale. Once Deliverable 3 exists for one operational area at one portfolio company, the same evidence pattern applies — unchanged — to every future Agentic AI deployment anywhere in the portfolio. ESG reporting, regulatory inquiries, board questions, and cross-border audits all become queries against the same kind of record. The evidence problem stops being a project each time. It becomes a query.
We will agree exact targets with AD Ports and the Strategy Office at kickoff. Our recommended starting set, all observable and verifiable:
Every measure is something the team can verify themselves.
Weeks 1–4 — Preparation. Sponsor and champion confirmed. Operational area chosen with AD Ports. Adjacency portco chosen (Aramex or Etihad Rail). Read-only connections established. Boundary policies drafted and reviewed. UAE-resident infrastructure topology and model-residency decisions completed.
Weeks 5–10 — First half of active operation. Deliverable 1 Agents go live in shadow mode alongside the AD Ports operations team — no production decision authority. The first cross-portfolio signal channel goes live by week 8. The evidence trail accumulates from day one. Weekly readouts to sponsor and champion, with course corrections in real time.
Weeks 11–15 — Second half of active operation. Second cross-portfolio signal type goes live. Case-pattern memory has grown enough to be visibly useful to the operations team. Mid-engagement readout to the Strategy Office — a structured demonstration of all three deliverables in use, with the data accumulated to date.
Weeks 16–17 — Validation and decision. Measures-of-success review with AD Ports and the Strategy Office. Compliance-question demonstration with full evidence export. Operator-experience review. Phase 2 commercial proposal delivered before week 17. Conversion decision by ADQ.
The engagement delivers. We meet or exceed the agreed measures of success. The natural next step is a Phase 2 engagement that extends the same pattern to a second portfolio company in the cluster, and begins the longer-running work for the Strategy Office on tracking ADQ's named investment theses — the Energy Capital Partners partnership, the Eni critical-minerals JV, the Limagrain seed development, and others. A Phase 2 commercial proposal is delivered before week 17 so there is no contracting gap.
The engagement partially delivers. We agree openly what worked, what did not, and whether to extend, repackage, or close out. ADQ owes nothing beyond the original engagement fee.
The engagement does not deliver. We close out cleanly. ADQ owes nothing beyond the original engagement fee. The work, the learning, and the evidence pack remain with ADQ — they have standing internal value even if the engagement does not continue.
The engagement is deliberately small. But every piece of it is built to scale, and each piece maps directly to one of ADQ's value-creation drivers:
The 120 days are the smallest defensible step toward portfolio-scale capability. Every next step builds on what this step puts in place.
Fixed-fee engagement for the 120 days. A commercial term sheet accompanies this document. Phase 2 — extension to additional portfolio companies, plus the Strategy Office thesis-tracking work — is structured on outcome-linked terms: payment tied to the value the work creates, not the time spent on it.
If the engagement does not deliver, ADQ owes the agreed engagement fee and nothing more.
We would rather be honest than impressive. We are deliberately starting small because the things we are describing — portfolio visibility, thesis tracking, evidence trails, working knowledge — are easier to claim than to deliver. We would rather earn the right to take on more, by doing the small work well, than promise a transformation we cannot defend in 120 days.
We also want to be clear about the partnership we are proposing. ADQ is the architect of its own value creation. What we bring is a way to make the connective work — between portfolio companies, between commitments and outcomes, between people and the knowledge they hold — easier and faster. The vision is yours. We are here to help make it operate.
We are happy to walk this through with the Strategy Office, with the Digital and Innovation team, or with AD Ports' executive committee — whichever order suits ADQ. We can be ready to begin within four weeks of a green light.