Agent commissioning and operations
We commission agents into your real systems, gate every action that matters behind a human approval, and stay on to run them afterwards. Before you talk to us, read a sample of the audit trail you would get.
Evidence
We started this practice in 2026. Rather than a wall of borrowed logos, here is the work itself. Every artefact below is a real output of the method, synthetic or redacted, and labelled as a sample. If you want to judge an engineering firm, judge its artefacts.
approval APR-2026-0731-004 agent invoice-triage action issue_credit_note amount EUR 4,820.00 policy irreversible, approval required requested 2026-07-31T09:14:22Z approved 2026-07-31T09:16:41Z by r.shah latency 2m 19s receipt sha256:9f2c1a...d47b
capability alone approve never read_ledger yes . . draft_reply yes . . send_customer_email . yes . issue_credit_note . yes . refund_over_10k . yes . delete_record . . yes change_own_permissions . . yes
suite invoice-triage / v14 cases 412 pass 397 96.4% fail 15 3.6% vs v13 -0.8pp regression, investigate p95 latency 1.84s cost / 100 $0.41 gate blocks merge below 95.0%
09:14:22 agent.plan classify, extract 09:14:23 tool.call ledger.read(88213) ok 09:14:24 policy.check issue_credit_note HOLD 09:14:24 approval.raise to r.shah (ipad) 09:16:41 approval.grant r.shah 09:16:41 tool.call ledger.credit(4820) ok 09:16:42 receipt.write sha256:9f2c1a...d47b
incident INC-0042 title eval regression after migration detected monitor: pass rate 96.4% to 91.2% action agent frozen, work to human queue cause provider changed a default sampling parameter without notice fix parameters pinned, gate added to CI elapsed 00:41 impact no customer-visible effect
agent/invoice-triage/ charter.md what it is for, and is not permissions.yaml the matrix, versioned evals/ 412 cases, run on change runbook.md what to do at 3am receipts/ every action, signed owner a named human being Yours, in your repository, whether or not we are still here.
The problem
Buyers are not short of tools. They are short of systems that survive contact with a real business and can explain themselves afterwards. The gap is not intelligence. It is operations.
| 10% | No more than one in ten organisations is scaling AI agents in any given business function, though 88% use AI somewhere. McKinsey, Nov 2025 |
| 60% | Nearly six in ten enterprises have seen no measurable improvement in total cost of ownership from deals that include agentic AI. BCG, Feb 2026 |
| 21% | Only one in five companies deploying agents reports a mature model for agent governance. Deloitte, Jan 2026 |
| 2 in 3 | Two-thirds of enterprises now expect their providers to build and operationalise their priority use cases, not just build them. BCG, Feb 2026 |
| 40% | Gartner predicts over 40% of agentic AI projects will be cancelled by the end of 2027, citing escalating costs, unclear business value, or inadequate risk controls. Gartner, Jun 2025 |
Services
Every engagement is tailored to the work in front of it, and scoped in writing before anybody starts. We will not sell you the second rung without the first, because we would only be guessing about what your agents actually do.
Two to three weeks
We put an eval harness on the agents you already have, find out why they fail, review the governance gaps, and write you a commissioning roadmap. If you paid for a pilot that stalled, this is the rung that tells you whether it can be saved, or whether it should not be.
Six to ten weeks
One agent, doing one job, inside your real systems. Integrations, evals, guardrails, human approval paths, a runbook, and acceptance criteria tied to a number you chose before we started. Not a demo. A thing that does the work, with a named human accountable for it.
Ongoing
We keep the watch. Monitoring, drift and eval regression, model migrations, and a monthly report in plain English about what your agents did, what they cost, and what we changed. This is the rung that decides whether the other two were worth anything in a year.
How we scope
Consulting here is tailored, not packaged. We look at what you have, agree in writing what we will do and what we will deliver, and give you a number before any work begins. That number does not move unless the scope does, and if the scope moves we say so at the time rather than in the invoice.
Method
Five stages. They are named for the work, and they are the same five whether the engagement is a two-week audit or a two-year retainer.
Measure before you move. We instrument what exists and take a baseline. You cannot improve a number nobody has ever taken, and you cannot prove you improved it either.
We write the permission matrix. What this agent may do alone, what needs a human to approve, and what it may never do under any circumstance. Clients tell us this is the document they did not know they needed.
Build, integrate, evaluate against the baseline. Guardrails and approval paths go in during the first week, not bolted on in the last. Every action that cannot be undone gets a named human in front of it.
We agreed a number at the start. Either it moved or it did not. Acceptance is not a demo and it is not a slide deck. If we missed, we say so, and the report says why.
Monitoring, regression, migration, and a monthly report. Agents drift. Providers change defaults without telling you. Somebody has to be watching, and it should be somebody who is paid to.
Every engagement leaves behind an agent file: the permission matrix, the eval suite, the runbook, the receipts, and the escalation path. It lives in your repository. It works whether or not we are still here. We think that is the minimum a consultant owes a client, and we are aware that it is not the industry norm.
Limits
A firm that will do anything has no method. These are not caveats buried in a contract. They are the shape of the practice.
No no-code automation piecework.
No model training or fine-tuning engagements.
No staff augmentation.
No hosting your inference on our hardware.
No agent takes an irreversible action without a named human approving it.
No borrowed credibility.
Regulation
A lot of people are currently selling against an August 2026 deadline for human oversight and logging. That deadline moved.
The Digital Omnibus was endorsed by the European Parliament on 16 June 2026 and given final approval by the Council on 29 June 2026. It defers the obligations for high-risk systems, which are the ones that concern anybody operating agents.
We mention it because a deadline is a poor reason to build an audit trail. A better reason is that you cannot operate what you cannot see.
| Obligation | Applies from |
|---|---|
| Article 50 transparency, Article 4 AI literacy, national enforcement and penalty powers | 2 Aug 2026 |
| Stand-alone high-risk systems (Annex III), including human oversight (Art. 14) and logging (Art. 12) | 2 Dec 2027 |
| High-risk AI embedded in regulated products (Annex I) | 2 Aug 2028 |
Your data
Class S / sensitive
Finance, legal, personal data, anything near a credential. Never enters a shared model context. Hosted use only under zero-retention terms with logging off, and tagged in the trace so you can see it was.
Class I / internal
Code, documents, operations. Hosted frontier models by default, because capability matters more here and the exposure is lower.
Class P / public
Marketing and outbound content. Cheapest capable route. It is reviewed by a human before anybody sees it anyway.
Per-client credential handles. Your data never crosses into another client's context, and it never enters ours.
Principals
A boutique is its people. Anonymous firms are not trustworthy, and a leveraged pyramid of analysts is not what you are buying here.
Rishiom Shah
Thirteen years building, operating and selling software. Runs a company whose daily work is carried out by AI employees under approval gates, with receipts for everything they do. Building Orbiter Dev, a governance layer that puts one approve button in front of every agent. The approvals land on an iPad beside the keyboard, which is where this practice started.
Aman Abhishek
Aman co-founded the practice with Rishiom, after a long time spent planning it between them. He moved back to India in 2026 and runs the practice from there. He owns the half of this work that a client only feels later: whether the agent still holds six months after handover, and whether somebody actually answers when the escalation path gets used at three in the morning.
Questions
It depends on what you have, and we will not pretend otherwise by putting a number on a page before we have seen it. What we will do is agree the scope and the number in writing before any work starts, and hold to it unless the scope changes. You will not discover the price at the end.
You get the founders themselves, and an audit that tells you honestly whether to proceed. If the answer is that your pilot cannot be saved, we will write that down, and you will have found out early and cheaply rather than late and expensively.
That is what the third rung is for, and it is why we agree it at signing rather than after something breaks. Agents drift. Providers change defaults without telling anyone. That is not a hypothetical, it is a Tuesday.
No. We have a default stack we know deeply. If you have a mandated platform, we work inside it. What we will not compromise on is the eval suite, the permission matrix and the approval path.
You are the deployer of record. Our job is to build the oversight that makes that position defensible: named human approvals on irreversible actions, and a logged trail of what was decided and by whom. Before you sign anything we will put professional indemnity cover appropriate to the engagement in place and show you the certificate, and we will send you the contract terms before you ask for them.
If the audit is not obviously worth it to you once we have described it, we are the wrong firm, and we will tell you that on the call rather than after it.
An audit, two to three weeks, and a written answer about whether your agents can be saved. If the answer is no, we will say so, and the roadmap will tell you what to do instead.