Robotiqa Advisory
Fixed-scope assessments and advisory retainers for institutions whose records carry legal or public weight.
Robotiqa is a single-principal practice. Engagements are led and delivered by the principal throughout — there is no team of juniors behind the proposal, and no incentive to extend scope. That constrains volume, which is the intended trade: the work suits institutions that want a senior second opinion on a consequential architectural decision, not a staffing supplement. Each of the offers below derives from published research — eight working papers, each archived with a permanent DOI on Zenodo, three of them also listed on SSRN — and comes with a defined deliverable. The series index is at Research.
Engagements are scoped for senior, focused input rather than sustained presence — assessments, reviews, diligence, and briefings.
AI Governance Readiness — OSFI E-23 / CBUAE
Problem. E-23 takes effect 1 May 2027 for every federally regulated financial institution, and OSFI expects visible progress during the transition. CBUAE's February 2026 guidance note places accountability for AI/ML with the board and senior management of every licensed financial institution. Both require a model inventory, risk tiering, and lifecycle governance defensible to an examiner.
Deliverable. Gap assessment against the applicable guideline, model inventory review, risk-tiering recommendation, prioritised remediation roadmap.
Self-assessment. The Decision Custody Check — ten questions on whether each automated decision has an owner, a record and an explanation.
Board & Executive Briefing
Problem. Boards are being made accountable for AI governance faster than they are being educated on it.
Deliverable. One prepared session on E-23, CBUAE, DIFC Regulation 10, or vendor concentration risk — built against your own estate, not delivered generically. Written summary follows.
Technology Due Diligence
Problem. An acquirer needs to know what is actually inside a target's technology estate — concentration risk, exit cost, licence exposure, architectural debt, AI and model liabilities — on a transaction clock.
Deliverable. Diligence report covering architecture, governance maturity, third-party concentration, post-merger integration risk.
Buyers. Private equity, corporate development, advising counsel.
Independent Architecture Review
Problem. A platform, vendor or migration decision is on the table and the organisation wants a senior read from someone with no stake in the outcome.
Deliverable. Written opinion with reasoning, risks, and dissent where warranted.
Exit-Capability Assessment
Problem. A platform that only one supplier can operate is a sovereignty exposure, not merely a commercial dependency. Most organisations have never priced their exit, and discover the number during a renegotiation — at the worst possible moment, with no leverage.
Who this is for. CIOs, CTOs, procurement and technology risk functions facing a major renewal, a consolidation decision, or a regulator asking about concentration risk.
Deliverable. An exit-capability register covering the critical platform estate; a measured exit cost per platform rather than an assertion; identification of the dependencies that cannot currently be exited at all; and a remediation roadmap sequenced by exposure.
Selected engagements
Examples from the practice, described by sector. Selected client names are published only where permission permits.
- A provincial government — grants disbursement
- A provincial government needed a grants-services platform on which every disbursement of public funds could be traced end to end, and which stayed available through peak intake. Robotiqa designed the multi-tier architecture: intake, adjudication and disbursement separated into their own tiers, the audit record each tier had to leave, and the availability each had to meet. Delivered as an architecture package the government's delivery partners built against. Nearest offer: Independent Architecture Review.
- A global bank — e-banking solution review, and agentic-AI guidance
- A global bank asked for an independent solution review of its e-banking platforms ahead of re-architecture. Robotiqa reviewed the designs as built, set out the risks and the dissenting view where the proposed direction did not hold, and produced the target architecture the programme then built towards. In a separate piece of work, Robotiqa wrote the bank's internal guidance for the governed use of agentic AI. Nearest offers: Independent Architecture Review; Board & Executive Briefing.
- A Canadian credit union — architecture method, then automation strategy
- A Canadian credit union had architects but no shared method. Robotiqa established a TOGAF-aligned enterprise-architecture methodology — the artefacts, the review gates and the governance forum they feed — so that platform decisions left a record the next team could follow. Later, the same institution asked whether to extend its robotic process automation. Robotiqa assessed the options and advised against: the recommendation was to stop investing in RPA and plan for AI, with no tool chosen — the point was what AI would reach, not which product to buy. Nearest offers: Independent Architecture Review; AI Governance Readiness.
- A property developer — capability model, maturity, roadmaps
- A property developer and asset manager with several lines of business wanted to know what its technology could and could not support before committing to a multi-year programme. Robotiqa modelled the business capabilities across every line of business, assessed the maturity of each, and produced the technology roadmaps the executive used to sequence investment. The same method is what the Technology Due Diligence offer applies to an acquisition target. Nearest offer: Technology Due Diligence.
What a written deliverable contains
The section structure of a written review — what it contains, in what order, and what each part is obliged to say — is set out at sanjeetkumar.com/work/studio (opens in a new tab).
Below is what the risk section of such a report looks like. It is illustrative: the rows are invented to show the columns a matrix has to carry, and are not findings about any organisation. Exposure is stated as consequence / likelihood, in that order — a rating whose basis is not written down cannot be argued with.
| Risk | Exposure | Current control | Gap | Recommended action |
|---|---|---|---|---|
| Single-vendor dependency in a core platform | Severe / low | Contractual SLA and support agreement | Contract governs availability, not departure; no priced or tested alternative | Commission an exit assessment; price the migration and set a re-test interval |
| No tested exit path for a hosted service | Severe / medium | Data-export feature exists and is documented | Export has never been run end to end, and nothing consumes the output | Run one full export into a standing target; record the elapsed time and the losses |
| Absent model inventory | Major / high | Procurement approval required for new tooling | Approval captures purchases, not models embedded in tools already bought | Establish an inventory with an owner per entry; make it a release gate |
| Decision records not retained beyond the system that made them | Major / high | Application audit logs retained per platform policy | Retention is tied to platform lifecycle; the obligation to explain outlives it | Separate decision records from the deciding system and retain to the obligation |
| Recovery tested for infrastructure only | Major / medium | Annual failover exercise, passed | Exercise proves the service restarts, not that the record is intact and complete | Add a record-integrity check, with a pass condition stated in advance |
| Reference data with no accountable owner | Moderate / medium | Change control on the systems that consume it | Control governs consumers, not the source; corrections propagate unevenly | Name an owner; publish the source of truth and the correction path |
Each row also carries an owner type — platform owner, service owner, data and AI governance lead, records and compliance owner, continuity owner, data domain owner. A working report names the individual; an illustrative one can only name the class. Either way the column is not optional: a risk with no owner has been recorded, not assigned.
Engagement models
- Fixed-scope assessment
- A defined question, a defined deliverable, a defined end date.
- Advisory retainer
- A standing senior sounding board for an architecture or governance programme, typically a small number of days per month.
- Board and executive briefing
- A single session on sovereignty, exit-capability, or AI accountability, prepared against your own estate rather than delivered generically.
Fees
Fees are quoted per engagement following an initial conversation, and depend on scope and access rather than headcount.
Not sure which of these fits?
Most engagements begin as an unstructured conversation about a decision that is already on the table. That conversation is free and carries no obligation.