Product Updates - Industry's most advanced Allocation Engine, 4 new integrations - Redis, Grafana, Vercel and ClickHouse

Most FinOps practitioners want allocation accuracy above 90% — measured, not estimated — before they trust a chargeback number enough to let it move real budget between teams. This release, the engineering team built a daily, automated way to prove that number, rather than simply claiming it.

What shipped

The allocation engine rebuild reached its cutover phase this release, re-pointing production reads to the new engine while a daily automated parity check compared legacy and next-generation output on frozen snapshots, to the cent, with known noise sources like demo organisations and expected rounding drift explicitly excluded so genuine discrepancies couldn’t hide behind them. Commitment-benefit distribution — historically one of the most disputed calculations in any chargeback practice — became an explicit, visible adjustment rather than an implicit formula. Lookup tables, reusable tag sets, and an expression-based rule type with a consolidation assistant gave FinOps practitioners direct tools for building custom allocation logic without needing a spreadsheet.

The notification platform’s pilot cutover reached every organisation on the platform, with anomaly detection and tagging-compliance digests registered as first-class event types, and granular unsubscribe controls letting a recipient turn off one event type rather than all of them. Five new or extended AI and infrastructure billing connectors shipped: OpenAI and Claude.ai Enterprise connection support, a ChatGPT Enterprise compliance feed, a Cursor billing connector, and Cloudflare billing, alongside a unified GitHub connect experience.

A round of platform consolidation landed alongside all of it too: a redundant backing database was retired entirely, a standalone MCP server was merged into the core backend, and the backend CI workflow was parallelised specifically to stop paying for idle runners — applying the same FinOps discipline internally that the platform applies to a customer’s own infrastructure bill.

Why parity testing beats trust

An allocation engine migration is only as credible as the proof behind it. Most FinOps practitioners set a specific, measurable bar before trusting a chargeback model at all — allocation accuracy above 90%, measured rather than estimated, with fewer than two disputed allocations per quarter as the practical sign that bar has actually been cleared. A vendor announcing a new allocation engine with no parity methodology is asking an organisation to take that 90% on faith.

This release’s daily parity check answers that gap directly. Rather than a one-time validation before cutover, the check runs continuously — comparing legacy and next-generation engines on frozen snapshots at cent-level granularity, every day, with demo organisations and expected rounding drift explicitly excluded so a real discrepancy can’t be dismissed as noise and noise can’t be mistaken for a real problem. Commitment-benefit distribution getting its own explicit chargeback adjustment matters for the same reason — discount-sharing disputes are disproportionately common in FinOps practice, and an opaque formula is exactly the kind of thing that pushes an organisation past the two-disputes-a-quarter line practitioners consider healthy.

Allocation Parity — What This Release Proved
──────────────────────────────────────────────────────────────
Check                          What it verifies
────────────────────────────    ──────────────────────────
Daily frozen-snapshot diff       Legacy vs. new engine match
                                   to the cent, every day
Noise exclusion                   Demo orgs and rounding drift
                                   don't mask real discrepancies
Commitment-benefit adjustment     Discount-sharing made explicit,
                                   not buried in a formula
──────────────────────────────────────────────────────────────
Target: allocation accuracy above 90%, proven daily —
not asserted once at cutover
──────────────────────────────────────────────────────────────

Notifications reach every organisation

This release, the notification platform reached every organisation — the point at which the feature stops being an experiment and becomes infrastructure a customer can actually depend on. That expansion happened with fatigue safeguards already built in from earlier phases: daily email caps, digest coalescing, and now granular per-event-type unsubscribe controls, measured against a documented industry baseline of more than 2,000 weekly alerts with only 3% ever requiring action. Reaching every organisation without reverting to that baseline is the real achievement; reaching every organisation with a naive, uncapped alert firehose would not have been.

What this means for evaluating any allocation engine

A few questions worth applying to any vendor here: is there a daily, automated parity check against the prior system, or just a migration announcement with no ongoing proof behind it? Is commitment-benefit distribution an explicit, visible adjustment, or buried inside an opaque formula — since discount-sharing disputes are disproportionately common and opacity here erodes trust fastest? Have notifications been proven to reach every organisation, or just a pilot subset? Has platform consolidation been applied to the vendor’s own stack, evidence they’ve tested their own discipline before recommending it to anyone else? And does connector coverage extend past hyperscaler billing into the AI and infrastructure tools teams are actually adopting independently, since AI spend increasingly arrives through exactly those channels rather than central procurement.

90% allocation accuracy is a claim any vendor can make. A daily parity check, proven to the cent against the system it replaced, is the difference between a claim and a fact.

See allocation accuracy proven this way, against your own estate. Book a 30-minute session and we’ll walk through how parity validation, commitment-benefit distribution, and proactive governance would apply to your actual chargeback model. Request a demo →

Related reading

• Product Updates - Parity-Tested and Production-Ready: DigiUsher’s Allocation Engine — August 24, 2026
• Product Updates - The AI Telemetry Pipeline Goes Live — August 10, 2026
• Product Updates - Budgets Re-engineered to Purpose-Driven — September 21, 2026
• Designing Cost Allocation Engines That Finance Trusts: A FinOps Practitioner’s Playbook — May 14, 2026

© 2026 DigiUsher. All rights reserved. Privacy Policy · Terms of Service · Trust Center