Field notes · pragmatic AI
Writing for people who ship AI —
not people who talk about it.
Deployment playbooks, compliance guides, operating-model thinking, and the ROI math we actually use with clients. Written by the team that has the conversations.
Value by Design: Why Most AI Fails and How We Build Differently
Most AI projects fail not because of bad models but because value was never designed into the system. DILR's approach places AI where the P&L moves — then measures it.
Read the playbook →More from the archive
NHS SBS £900m AI framework: voice AI supplier guide
NHS SBS AI framework voice AI suppliers face the 23 June 2026 deadline. Clinical safety, DTAC, ISO 27001 and ambient voice registry — checklist inside.
Voice AI Buyer Leverage in 2026 Funding Cycle
Voice AI buyer leverage peaks in 90-180 day vendor funding cycle windows. Five contract clauses that move now after Vapi, PolyAI and ElevenLabs 2026 rounds.
Voice agent latency benchmarks: enterprise reality
Voice agent latency benchmarks enterprise buyers should demand — how accents, noise, interruptions and domain vocabulary break vendor demo-room numbers.
AI voice higher education: admissions enquiry guide
AI voice for higher education admissions answers every Clearing enquiry call on first ring, captures structured CRM data, and converts the surge to intake.
EC Article 50 guidelines: a voice AI deployer checklist
EC Article 50 draft guidelines divide voice AI transparency duties between provider and deployer — close the deployer-owned gaps before 2 August 2026.
AI voice platform: enterprise selection criteria
AI voice platform selection for enterprise buyers: the four procurement gates — integration, escalation, analytics, compliance — that decide Year 2 ROI.
AI voice agent QA and testing: an enterprise framework
AI voice agent QA testing for enterprise: the four-layer framework — unit tests, red-teaming, shadow calls, and 100% live production scoring at scale.
EU AI Act omnibus: what is delayed, what is not
EU AI Act omnibus delayed Annex III to 2 December 2027 — but Article 50 voice AI transparency still hits 2 December 2026. What enterprises must do now.
AI voice hospitality reservations: enterprise guide
AI voice hospitality reservations recover the 20-30% of calls UK hotels lose to hold time and after-hours. The revenue maths, the architecture, the rollout.
LLM vs scripted voice agents: enterprise guide
LLM voice agents vs scripted flow builders: which architecture survives enterprise production in regulated UK industries. The honest answer is hybrid.
AI voice governance: enterprise framework guide
Enterprise AI voice governance is the gate before procurement: the framework, artefacts, and EU AI Act deadlines every enterprise board now asks about.
PolyAI Canada Expansion: What Voice AI Buyers Should Read
PolyAI Canada expansion enterprise voice AI: read the May 2026 Toronto hub and 200+ customer signal as a buyer — three scaling models, one contract.
DNC Logic in AI Voice Diallers: Compliance Built In
DNC compliance for AI voice diallers must be infrastructure, not a feature. The architecture UK and US enterprises need to avoid ICO and FCC enforcement.
AI Voice Program Design: From Pilot to Enterprise Scale
AI voice program design for enterprise scale: the architectural choices at pilot stage that decide whether you reach 50,000 calls a month — and EBIT now.
MHRA AI Airlock: NHS ambient voice procurement
MHRA AI Airlock now gates NHS ambient voice procurement: the four architecture gates UK suppliers must clear before framework selection closes in 2026.
Voice AI data retention: enterprise GDPR guide
Voice AI data retention under GDPR: where call recordings live after hang-up, retention windows, lawful basis, and the architecture procurement now demands.
Voice AI Procurement in May 2026: Reading the Vendor Map
Voice AI procurement framework 2026: read Vapi's $500M, PolyAI's 200 enterprise customers, and Article 50 in a single four-axis vendor map before signing.
FCA AI response: voice AI financial services 2026
FCA Treasury Committee AI response now requires evidenced human-in-the-loop, bias testing, and concentration risk rationale for UK voice AI deployments.
Vapi, Amazon Ring, and the Enterprise Voice AI Bake-Off
Vapi Amazon Ring enterprise voice AI selection: the four-gate procurement methodology that beat 40 vendors in two weeks — what UK buyers should now copy.
Voice AI Valuations 2026: Funding Signals for Buyers
Voice AI valuation enterprise procurement 2026 — Vapi $500M, ElevenLabs $11B, PolyAI $750M, Decagon $4.5B. Read the funding map as a vendor risk score.
AI tool inventory: what ICO, FCA, EU AI Act require
AI tool inventory enterprise compliance is the first thing ICO, FCA and EU AI Act ask for. Build the regulator-ready list in 30 days — template inside.
Voice AI accuracy: beyond WER for enterprise
How to evaluate voice AI accuracy beyond WER: the four operational metrics enterprise buyers must score before any production deployment decision is signed.
AI voice for SaaS customer success: scale CSMs
AI voice for SaaS customer success automates the check-in layer, freeing CSMs for expansion. NRR-defending economics for FY26 boards under cost pressure.
AI Voice Operating Model: In-House vs Vendor vs Hybrid
AI voice operating model enterprise guide: in-house vs vendor vs hybrid TCO, time-to-deploy, engineering load, FCA risk. See where the £ actually lands.
Voice AI Vendor Consolidation: De-Risking the Buyer Stack
Voice AI vendor consolidation is now an enterprise procurement risk. The architectural defence: portable prompts, swappable telephony, neutral analytics.
AI outbound calling: GDPR and PECR compliance guide
AI outbound calling under GDPR and PECR: UK enforcement is rising, the £500k cap is gone, and the new £17.5m ceiling now hits non-compliant programmes.
AI voice logistics: dispatch calls without missed loads
AI voice logistics dispatch cuts cost-per-call below £0.50 and recovers a £1.6bn UK failed-delivery bill. Map four call types to AI containment that scales.
AI voice ROI: total programme economics
AI voice ROI for enterprise goes beyond cost-per-call. Model implementation, attrition cost, expansion revenue, and the five-line P&L your CFO will sign.
Real-time transcription AI voice: enterprise data layer
Real-time transcription AI voice calls is the data layer every downstream voice AI capability rides on. WER, confidence, and procurement gates inside.
AI voice insurance claims: the FNOL playbook
AI voice insurance claims intake at carrier scale: how UK insurers cut FNOL from 22 to 6 minutes under FCA Consumer Duty — the practical playbook inside.
AI voice pilot purgatory: Why 70% of programmes stall
AI voice pilot purgatory traps 70% of enterprise programmes. Four pre-pilot decisions on metrics, integration and exit criteria decide if scale happens.
AI voice utilities: cut call costs without breaking Ofgem
AI voice utilities convert meter-read, tariff and balance calls without breaching Ofgem Consumer Standards — if vulnerability detection is built in day one.
Voice AI hallucination: a procurement gate
Voice AI hallucination is now a regulatory event under SB 942, EU AI Act Article 50, and ICO rules — make containment a procurement gate, not a demo Q.
Voice AI Agent Quality Scoring: Automated QA at Scale
Voice AI agent quality scoring is the enterprise procurement gate of 2026. Score every call automatically, pass audit, deploy at scale with confidence.
HIPAA-Grade Voice Automation: What Healthcare Teams Need
HIPAA voice automation is more than encryption. PHI access control, audit trails, BAAs, and minimum necessary use decide if your healthcare deployment ships.
AI voice real estate lead qualification at scale
AI voice real estate lead qualification closes the gap between enquiry and first contact — where ~40% of UK property deals are quietly lost. Read more.
Voice AI Barge-In Handling: Why Interruptions Break Deals
Voice AI barge-in handling is a procurement criterion, not a UX detail. The benchmarks, vendor questions, and tier model that protect containment ROI now.
AI voice program KPIs: the enterprise guide
AI voice program KPIs go beyond completion rate. Track these nine outcome, economic, quality and risk metrics to lift payback inside six months — start now.
ICO AI Code of Practice: Voice AI obligations from May 2026
ICO AI Code of Practice (SI 2026/425) takes effect 12 May 2026. What UK enterprise voice AI must now evidence: disclosure, explainability, bias, redress.
AI voice fintech collections: ROI and FCA compliance
AI voice fintech collections cuts cost per contact 8× and gives FCA Consumer Duty audit trails on every call. The UK enterprise ROI and compliance playbook.
AI voice sentiment analysis: what enterprises measure
AI voice sentiment analysis on 100% of calls only pays off when the loop closes. The signals enterprises actually measure and a 90-day framework that ships.
Change management AI voice: what teams get wrong
Change management AI voice deployment fails on human factors, not tech. The 4 failure modes, the 90-day rollout, and the operating model that sticks now.
FCA AI Governance 2026: What Voice AI Deployments Must Do
FCA AI governance for voice AI: from 1 September 2026, UK voice deployments fall under SM&CR, Consumer Duty and the new Code of Conduct. Get audit-ready.
AI voice outbound enterprise sales: lead follow-up at scale
AI voice outbound enterprise sales fills the first three contact attempts SDRs skip — where 93% of conversions live. Capacity gain, not SDR replacement.
Voice AI orchestration vs platform: the enterprise choice
Voice AI orchestration vs platform decides whether you operate raw infrastructure or buy outcomes. The enterprise framework, the numbers, and the call.
Voice AI TCO: the hidden enterprise costs vendors hide
Voice AI TCO: Vapi $0.05/min sticker hides $0.30+ true cost. The enterprise procurement model finance teams use to compare orchestrator vs platform now.