Signal Pod — Validate
Most enterprise AI bets die in PowerPoint. The Signal Pod compresses thesis-to-evidence into 15 working days: opportunity teardown on day one, working agentic prototype by week two, executive readout with measured KPI deltas before Friday-of-week-three. You leave with go/no-go clarity and a refund clause if we miss.
Inside the pod
- Opportunity teardown (day 1)
- Live agentic prototype (week 2)
- Outcome scorecard with KPI baseline
- Executive readout deck + Loom
- Go / no-go clarity memo
- Principal (revenue strategy)60%
- AI engineer (full-stack)100%
- Designer (product / pitch)60%
- Growth analyst40%
# Signal Pod · Day-15 Readout
hypothesis: "Renewals copilot lifts gross-renewal NRR by ≥ 4 pts."
baseline_nrr: 104.2 %
prototype_nrr: 108.7 % (Δ + 4.5 pts, p95 < 0.03)
eval_pass_rate: 0.94 / 1.00
cost_per_run: $ 0.041
recommendation: GO → Scale via Compounder Pod (P2) starting wk 17
refund_line: not tripped3 weeks · fixed
- 1Week 1Opportunity teardown
Workflow shadowed end-to-end, KPI baseline captured, hypothesis sharpened, eval harness scaffolded.
- 2Week 2Prototype build
Working agentic prototype on real data, deterministic fallbacks, signed tool calls, telemetry on every run.
- 3Week 3Evidence + readout
Measured KPI deltas, executive readout, go/no-go memo, refund-line trip-wire reviewed publicly.
A cadence the CFO can audit
- Day 0Kickoff
Hypothesis lock-in, KPI baseline, success/refund line agreed in writing.
- Day 5Architecture review
Prototype skeleton, model routing plan, eval harness wired.
- Day 10Mid-sprint demo
Live prototype against synthetic + sampled real data.
- Day 15Executive readout
Measured KPI delta, go/no-go memo, scale-path proposal.
Everything the pod ships
- Hypothesis lock-in memoOne-pager fixing scope, KPI baseline, success line and refund trip-wire.
- Working prototypeDeployed in your sandbox or ours; source delivered under MSA.
- Eval scorecardQuality, cost, latency and policy evals — measured, not estimated.
- Executive readout deck + Loom30-slide narrative deck + 15-minute Loom for board distribution.
- Go / no-go memoRecommendation with three-path scale plan and budget envelope.
Skin in the game
If no working prototype is shipped by day 15, the £24k is refunded in full.
All source, prompts, evals and telemetry transferred under your MSA on day 15.
Your data, your region, your keys. Nothing leaves your tenant without written consent.
- ✓ C-suite betting on a single high-stakes AI thesis
- ✓ Pre-board approvals needing measured evidence
- ✓ Replacing a 6-month consulting discovery
- ✗ Open-ended R&D with no KPI
- ✗ Compliance-only programmes (use the Trustworthy AI Pod)
Killed a £4.2M agentic claims bet in 15 days · saved 9 months of build
"The Signal Pod gave us a no-go we trusted. The refund clause meant the analysis was honest, not consultative."
What buyers actually ask
That is a successful sprint. You get a measured kill-decision in 15 days instead of a soft death over 18 months.
Yes — preferred, in fact. We bring the runtime; your data never leaves your tenant.
Yes. All source, prompts, evals, traces and telemetry transfer to you on day 15 under your MSA.
If go, the same principal continues into the Compounder Pod (P2) with no re-discovery cost.
Plug Signal Pod into your P&L.
We'll respond within one business day with a scoping note, a fixed-price outcome contract, and a named principal cleared for your domain.
- • Fixed price · £24k / 3-week sprint.
- • Minimum term · 3 weeks · fixed.
- • Exit clause · Full refund if no working prototype is shipped by day 15.