Five outcome-bundled service lines: how we organise engineering work

A productized engineering firm needs a small, legible product catalogue. Too many services and buyers can’t tell what they’re buying; too few and you can’t fit the actual demand. We landed on five service lines, each a fixed-scope container with a defined entry point and a defined deliverable.

This post is the catalogue, plus the logic behind why each line exists.

1. Build

What it is. Net-new web, mobile, or data products built from scratch. Typical engagements run 12 weeks at a fixed price, milestone-based, with the deliverable code in your GitHub from day one.

Who it fits. Founders shipping a v1, established companies launching a new product line, or teams replacing a legacy internal tool with something modern. The defining feature is that the scope is knowable upfront.

Bundles inside Build: Web SaaS, mobile, marketplace, automation backend, regulated (HIPAA / DPDP / FCA-aligned).

2. Upstream

What it is. Direct replacements for over-priced enterprise SaaS. We rebuild the workflow on direct AWS / Anthropic / open-source billing, often at <5% of the per-seat licence cost. The rebuild is scoped against the SaaS spend it replaces; ROI lands in 6–12 months.

Who it fits. Companies paying $5K–$50K/mo per tool for software their team uses 15–25% of, or enterprises renegotiating $200K–$700K/yr renewals where the per-seat math has stopped making sense.

Catalogue: Harvey AI, Datadog, Salesforce, Klaviyo, HubSpot, Workday, Mixpanel, Segment, Shopify Plus, NetSuite, Zendesk, Intercom, Sentry, Pardot, SFMC, Notion Enterprise, Outreach, Yotpo, Jasper, Mailchimp — and 12 more.

3. AI Pod

What it is. A small embedded AI engineering team running 12-week sprints at a fixed monthly fee. Used by companies that need to ship production AI but don’t want to recruit, manage, or carry the FTE cost of a permanent ML org.

Who it fits. Non-AI-native organisations (finance, healthcare, legal, FMCG) that have one or two well-defined AI initiatives. Also high-growth tech companies that need to compress the ramp on a new AI product.

Typical outcomes: A production-deployed AI feature, a custom-fine-tuned model, an LLM-powered internal tool, or a research-to-prod pipeline.

4. Agents

What it is. Production LLM agents that replace brittle integrations or repetitive operational workflows. Typical engagement: 6–10 weeks fixed scope, including evaluation harness, observability, and rollback procedures.

Who it fits. Companies whose ops, sales, or service team is doing high-volume manual work that has been classified as “automatable but no integration exists.” Agents become economically viable when the integration cost exceeds the workflow value.

What we don’t do: Speculative “let’s see what AI can do” research. Agents are a delivery line, not an experimentation line. The use case is defined before we start.

5. Strategy

What it is. Stack audits, technical briefs, build-vs-buy memos. Fixed-price 2–4 week engagements that produce a written artefact you can take to a board, a CFO, or a different vendor.

Who it fits. Companies considering a major refactor, replacement, or build decision and who want a senior, vendor-neutral assessment before committing capital.

Outputs: A 15–40 page written brief with: current-state assessment, options evaluated, recommended path, rough effort estimate, and explicit risks. The brief is yours; you’re free to use it with us, with another vendor, or shelve it.

Why exactly five

Each line corresponds to a distinct buyer intent:

Buyer intent	Line
“I need to build something new”	Build
“I’m paying too much for this SaaS tool”	Upstream
“I need to ship AI but don’t want to hire a team”	AI Pod
“This manual workflow needs to go”	Agents
“I need a written brief before I commit”	Strategy

Each line has a different entry point on the site, a different SOW template, and a different default team composition. The discipline of keeping the catalogue small means buyers can self-route quickly. The discipline of making each line container-shaped (not task-shaped) means we can reshape an inbound enquiry into the right line without bait-and-switch.

What’s not on the list

Hourly engineering. Retainer-without-deliverable. “Discovery phases” that don’t produce a written artefact. Marketing campaign management. Staff augmentation under the label of “AI Pod.” We’ve turned away enough of these to know that the catalogue’s clarity is its strength.

---

Browse the catalogue: /build/ · /upstream/ · /ai-pod/ · /agents/ · /strategy/