Why 4 weeks specifically?
Four weeks is the smallest cycle that contains a full ship-to-production loop with proper engineering - eval suite, observability, gradual rollout, transition. Compress to two weeks and you skip the eval suite or the transition; expand to twelve and you accumulate spec drift faster than you ship.
DORA's research [3] shows elite engineering teams deploy multiple times per day with cycle times under a week. The Lightning Pod borrows this discipline and applies it to a single bounded scope: by week 2, the team is shipping daily; by week 4, the customer team owns the deploy cadence. The 4-week container is engineering-shaped, not consulting-shaped.
Pod composition: who's in the room
A typical Lightning Pod is five people: two senior product engineers (full-stack, your stack), one staff engineer for orchestration and architecture, one product or operations lead (paired with your operational owner), and one part-time security or compliance partner for regulated workloads. We tune by scope - heavier on data engineering for retrieval-heavy work, heavier on platform engineering for multi-agent shifts.
The customer-side counterpart matters as much as the pod composition. We require a named operational owner who reviews work weekly, an engineering counterpart who pairs daily, and a stakeholder reviewer who runs the demo each Friday. Without these three roles, the pod can't transfer ownership in 4 weeks - and we won't take the engagement.
| Role | Pod side | Customer side | Time commitment |
|---|---|---|---|
| Senior product engineer (×2) | Full-time | Engineering counterpart (1×) | Daily pairing |
| Staff engineer | Full-time | Tech-lead reviewer | 2x/week |
| Product / ops lead | Full-time | Operational owner | Daily |
| Security / compliance | Part-time (regulated only) | Security counterpart | 1x/week |
| Stakeholder reviewer | - | Decision-maker | Friday demo (1h) |
The 4-week shape, week by week
| Week | Engineering | Customer team |
|---|---|---|
| Week 1 | Eval suite (50+ cases) committed; OTel telemetry wired; first agent prototype | Operational owner reviews evals; provides golden traces |
| Week 2 | First deploy behind feature flag; bandit-routed canary; eval-gated CI live | Stakeholders use the feature in shadow mode; submit edge cases |
| Week 3 | Canary expanded to 10–25%; reviewer queues live; runbooks drafted | Customer team co-owns review queue; on-call shadowing |
| Week 4 | Full rollout; runbooks finalized; transition session | Customer team takes operational ownership |
View data table· Source: Aggregate Techimax engagement telemetry, 2024–2026
| Series | % pass rate |
|---|---|
| Day 1 | 42 |
| Day 4 | 64 |
| Day 8 | 78 |
| Day 14 | 88 |
| Day 20 | 93 |
| Day 26 | 96 |
| Day 30 | 97 |
Want a Lightning Pod for your roadmap?
Senior engineer replies within 24 hours. Free working session to scope.
Where it fits and where it doesn't
- Best fit: a single AI feature with a measurable business outcome and a named operational owner.
- Good fit: AI Rescue scopes (hardening an existing copilot in 4 weeks).
- Marginal fit: multi-feature programs - better served by a longer-running embedded engagement (8–16 weeks).
- Not a fit: scopes that haven't passed our POC checklist - the underlying business outcome must be clear before week 1.
What gets shipped: typical week-4 deliverables
We track four shipped artifacts at engagement close: a production agent serving real traffic, a checked-in eval suite of 50–120 cases gating CI, OpenTelemetry traces flowing into the customer's APM [2], and runbooks for the top six failure modes. Anything missing from this list means we extended into a fifth week - which we have, occasionally, but the goal is week-4 close.
Median outcomes across 30+ Lightning Pods: 11 weeks of cycle compression vs in-house build, 3.4× lower all-in cost, eval pass-rate ≥ 92% at handoff, stakeholder weekly active rate ≥ 75% at week 8. We share the full benchmark on request.
View data table· Source: Techimax engagement telemetry 2024–2026
| Series | see metric |
|---|---|
| Cycle time (weeks) | 30 |
| Eval pass-rate at handoff (%) | 92 |
| Cost vs in-house (% of) | 30 |
| Stakeholder usage week 8 (%) | 81 |
| Pod weeks to first deploy | 2 |
The Lightning Pod is engineering-shaped, not consulting-shaped. Four weeks contains a full ship-to-production loop with proper rituals - anything less skips a step; anything more accumulates drift.
What happens after week 4
We don't disappear at week 4. The transition includes a 30-day post-engagement support window where the pod is on-call for questions, runbook updates, and the first incident review. Total commitment: roughly 4 hours/week from one pod engineer for 30 days, included in the engagement.
By day 60 the customer team is operating independently. We measure success by the absence of urgent calls - a quiet 30-day window means the rituals stuck. We follow up at the 90-day mark with a brief retrospective and offer (rarely needed) a tune-up engagement if drift has accumulated.
References
- [1]Forward-deployed engineering at scale - Techimax engineering research (2025)
- [2]OpenTelemetry GenAI semantic conventions - OpenTelemetry SIG (2025)
- [3]DORA 2024 State of DevOps Report - Google Cloud / DORA (2024)
- [4]Team Topologies - Skelton & Pais (IT Revolution) (2024)
- [5]The state of AI in 2025: Agents, productivity, and risk - McKinsey & Company (2025)
Frequently asked questions
What if our scope is bigger than 4 weeks?
We run 8-week engagements ("Velocity Pods") for multi-feature scope. Same shape, longer cycle. Bigger than 8 weeks usually means you want a longer-running embedded engagement; we run those too.
How is this priced?
Fixed-fee per pod-week. Pricing scoped at the kickoff session based on team size and scope. We share the rate sheet on request - no SDR ladder.
Does the customer team need AI experience?
No. Lightning Pods are designed to bring the rituals to teams without prior AI engineering experience. By week 4 the customer team can run the system.
What does the kickoff process look like?
Free 60-min scoping call with a senior engineer to validate scope fit. If it's a fit, a written scope and rate sheet within 48 hours. Kickoff happens 1–2 weeks after signature; we typically have pod availability within 30 days for non-cleared work and 60 days for cleared engagements.
Can a Lightning Pod work fully remote?
Yes. Standard format is 1–2 days onsite in week 1 and week 4, remote with daily customer-channel presence the rest of the time. Fully remote also works for distributed teams; the in-room intensity is replaced with persistent Slack and shared tooling.
What if the operational owner can't commit daily time?
Then a Lightning Pod won't transfer ownership and we'll propose a different shape. The daily presence of the operational owner is non-negotiable - we've measured the failure rate of engagements that compromise on this and it's high enough that we'd rather decline than ship a hollow handoff.
Do you offer a money-back guarantee?
We offer a free week-1 scope confirmation: if after the first week's gap report the customer decides the scope isn't a fit, we walk away with no charge for the week. We've never had to invoke this - but the option means scope-fit is genuinely the customer's call.