Four commitments hold every engagement: the price ties to the deliverable, the pod absorbs delivery risk, composition — senior experts plus orchestrated agent capacity — is declared in the proposal, and the pod-day is the unit.
customer-portal · day 5
07:42
Commits
14
Tests passing
286 / 286
Deliverables accepted
8 / 13
Pod-days to date
5.0
staging.customer-portal
Indicative daily artefact. Every engagement produces one before standup, every working morning.
What is in hereSenior experts on every job
1–3experts
Each pod is 1–3 senior experts directing purpose-built agents. Composition stated in every proposal.
Proposal back the next day
24hours
A self-driven assessment turns into a fixed-price proposal — outcome pod or day-rate pod — by the next business day.
Live in weeks, not quarters
3–8weeks
Daily evidence from the first business day. You see the work, not status updates.
The unit of delivery
A pod — one to three senior experts plus orchestrated agent capacity, sized to the engagement. You buy what the pod produces, not how many hours each individual logs.
What you receive
Code in your repository, staging environments you can open, tests that pass, and documentation produced alongside the build. A fourteen-day warranty on accepted deliverables.
Cadence
Most engagements complete in three to eight weeks. Daily evidence is verifiable on the morning after. Weekly scope reviews are available; never required.
Commercial model
Day-rate pod: a fixed pod-day rate against a defined composition — one to three senior experts with orchestrated agent capacity. Outcome pod: a fixed price for a defined deliverable by a defined date; the pod bears the delivery risk. Stop a day-rate pod with one business day of notice; pause briefly when needed. Specific pricing is stated in the proposal.
Capabilities
Each platform is led by a senior expert with depth on it. Inside every engagement, human judgement directs orchestrated agent capacity working as one pod — the expert sets the bar; the agents carry the throughput.
Production builds on the Anthropic and OpenAI stacks: custom agents, Claude Code skills, Custom GPTs and ChatGPT Apps, Realtime voice, retrieval pipelines. Substrate matched to the engagement.
TypeScript-first delivery for web apps, integrations, automations, and migrations — built on a modern stack your team can operate.
ITSM, CSM, HRSD, and ITOM extensions, Flow Designer, Integration Hub, Service Portal, and scoped applications. Panel-friendly delivery for regulated industries.
Power Platform, Azure, M365 and Teams customisation, Dataverse, and Copilot Studio. Significant government and enterprise estate.
Sales Cloud, Service Cloud, and Experience Cloud extension work — flows, Apex, LWC, integration. The newest discipline, ramping in line with confirmed client demand.
Standalone test engagements — automated regression suites, test infrastructure standup, coverage audits, post-incident hardening. Runs as its own pod when the work is test-shaped, and as a Capability inside every other discipline by default.
What we deliver
Most engagements look like one of the four bespoke shapes on the left or one of the accelerators on the right. The shape is named, sized, and sequenced before work begins.
Bespoke shapes
Scoped to your context
Authenticated client surfaces with billing, document access, and self-service flows. Replaces or extends a Salesforce or ServiceNow front door.
Salesforce ↔ SAP, ServiceNow ↔ Workday, custom event pipelines. Idempotent, observable, and explicitly tested under failure.
Schema-aware moves with parallel-run validation, reconciled balances, and a signed cutover. Rollback is rehearsed before it is needed.
Decision-grade dashboards on top of warehouse, OLTP, or hybrid sources. Definitions captured against the metric, not the chart.
Accelerators
Full catalogueProduction-ready AI feature inside your existing product — spec to shipped, with tests and observability.
Internal data and tooling wired to Claude via the Model Context Protocol. AI that knows your stack.
Identify, instrument, and fix the biggest drag on your team — build times, test flakiness, tooling gaps.
A legacy system moved to a modern architecture — incrementally, with tests at every step.
An evaluation harness and targeted safety audit. You learn the failure modes before users do.
A sequenced AI adoption roadmap grounded in what you actually have — not what AI can do in theory.
The artefact
The hero shows day five. Below is the same project at day eighteen — at handoff. The shape never changes; the scope it evidences does.
customer-portal · day 18
07:38
Commits
6
Tests passing
412 / 412
Deliverables accepted
13 / 13
Pod-days to date
18.0
cutover.customer-portal
How a pod operates
Four phases, each with a measurable artefact. Decisions are recorded against the deliverable. Status emerges from the work itself.
A self-driven scoping session converts your request into a delivery brief: systems in scope, constraints, risks, and an explicit definition of done.
A pod — one to three senior experts plus orchestrated agent capacity — works on your outcome only. Pod composition and pricing are stated before work begins.
Every working morning, before standup: committed code, passing tests, a live staging environment, and a written record of every decision. You read the work itself.
Accepted work transfers with code, runbooks, tests, and full repository history. You operate what was built without ongoing dependency on us.
Engagement shapes
Same talent. Same orchestration. Same delivery economics. The difference is procurement — fixed price for a defined deliverable, or a defined day rate for continuous output.
Fixed term
Fixed price for a defined deliverable by a defined date. Used when scope is definable upfront and the buyer wants total cost certainty. The firm bears the delivery risk.
Best for
Portals, integrations, migrations, dashboards — work whose shape is knowable before it begins.
Fixed-term or ongoing
Priced per pod-day with defined composition. Used when scope is genuinely fluid — ongoing operations, embedded engineering, exploratory product work, modernisation programmes whose shape evolves.
Best for
Continuous output where the work direction shifts faster than a contract change-order can keep up.
Discovery pod · 3–5 days, day-rate priced, onsite or remote
Optional bridge when scope genuinely needs scoping. Output is a proposal for the main engagement. We resist anything longer.
The alternatives
A pod sells the outcome — or a fixed pod-day — and absorbs the delivery variance. The two columns to weigh against any path on the table.
Comparison is illustrative; day-rate ranges and engagement shape are stated in the proposal.
Governance
The engineering substrate is new. The acceptance bar is the same one your CIO already trusts. Six commitments hold across every engagement.
Cloud regions chosen explicitly against the engagement. Residency captured in scope. No silent placement decisions.
A named expert signs off the pod’s output. The acceptance bar does not move with the substrate.
Repositories transfer to your GitHub on handoff. Code, history, and runbooks come with you.
No model training on client code, conversations, or context. Disclosed in the Terms of Service.
$1M–$2M professional indemnity in place from the first engagement. Liability capped at fees paid.
Tests, security scan, expert review, named acceptance — required before any deliverable ships.
Begin
Free. No commitment. No sales call. The assessment is self-driven — pod composition, scope, and price land on the next business morning.
What you will receive