# Battlecard — Verixa vs AmpleLogic (INTERNAL / reactive use) **Use:** objection handling + Lane B accounts already on an incumbent. **NOT** a slide in the first-send Lane A deck. **Claim discipline:** Verixa = *implemented · internally tested · validation-ready · in customer-validation phase*. NEVER *validated / compliant / audit-ready / full-suite replacement*. AmpleLogic column = their public marketing claims, not independently verified. --- ## Clean side-by-side | | **AmpleLogic** | **Verixa** | |---|---|---| | **What it is** | Mature low-code "aPaaS for Life Sciences" — broad deployed suite | AI-native GxP eQMS, governance-first | | **Best fit** | Enterprises consolidating 5–10 legacy systems | No-mature-eQMS accounts wanting one governed AI evidence workflow | | **Module breadth** | 14+ deployed modules | ~70 modules built (QMS core + APQR, batch, stability, OOS/OOT, inspection-readiness) | | **Module gaps (repo-verified)** | ✅ Dedicated LIMS · ELN · QC Planning · eLogbook · CAPS · CPV · CVS | ✗ Verixa lacks these **7** (lab + equipment + specialised validation) — but **has** DMS, LMS, batch records, APQR, stability, UAM, regulatory (so most of the *quality* footprint matches) | | **AI approach** | AI **bolt-on** to pre-AI core — recommendation engines, agentic report builder | **AI-native** — evidence/exception engines, predictive scoring, MIRA copilot as primitives | | **AI governance** | "AI-powered"; no published Annex 22 / EU AI Act framework | Architected to **EU Annex 22 (draft) / EU AI Act high-risk** — intended-use, model registry, HITL gates, advisory↔system-of-record split | | **Audit-trail integrity** | Configurable audit trail (standard) | **SHA-256 hash-chained, append-only, tamper-evident** (`verifyHashChain`) | | **Validation status** | Markets modules as *pre-validated / validation-ready* but uses absolutes freely — "GAMP 5 **Validated**" (Stability), "**validated** for pharma" (LMS), "100% FDA **Compliant** / **audit-ready**" (eQMS). **Customer still owns validation regardless.** | **Validation-ready · IQ/OQ/PQ authored · in customer-validation phase**; customer executes & approves. *Same actual posture, stated honestly — honesty edge to us.* | | **Proof** | 120+ customers · 500+ implementations · 15+ yrs · ISO 9001/27001 · SOC 2 · G2 4.8 | Pre-reference · founding-partner cohort forming | | **Deployment** | "Go live in weeks," consultant-led, full suite | 90-day bounded design-partner sprint · 1 facility · 1 workflow | | **Pricing (India)** | Not published — prior intel ~₹10L Y1 entry (verify) | Founding-partner: Phase 0 ₹3L · Phase 1 ₹9L | --- ## Objection handling (say it this way) **"AmpleLogic has 120+ customers — why you?"** > "They're the broad, proven incumbent suite — correct. We're not competing on breadth or vintage. We're a focused, *governed* AI evidence workflow for one bounded GMP process, architected to the AI rules — Annex 22, EU AI Act — that every QMS will have to meet. Phase 1 proves it on your data in 90 days, and you own validation." **"They're already validated / Part 11."** > Never contest their claim or echo "validated." → "We're validation-ready and hand you the IQ/OQ/PQ package to execute with your QA — you approve validation before any production use. We'd rather be precise about that than overclaim." **"They do AI too."** > "They added AI recommendation engines on top of a pre-AI core. We built the governance layer first — intended-use statements, human-in-the-loop gates, an advisory-vs-system-of-record boundary, and a tamper-evident hash-chained audit trail. Governed AI, not 'AI-powered.'" **"Is Verixa just a point tool?"** > "No — the platform spans the full QMS plus APQR, batch, stability, OOS/OOT and inspection-readiness, ~70 modules built. We *start* you on one bounded workflow on purpose, so value is provable fast — then expand." --- ## TCO / sovereign-AI (reactive — don't lead with this) **"Their open-source stack saves up to 70% on infrastructure."** > "We're on the same modern open-source foundations they tout — PostgreSQL, containers, Kubernetes — but as cloud SaaS you don't run any of it: nothing to license, host, patch, or migrate. Their infra-savings story is for a big-pharma IT shop replacing a legacy Windows/SQL Server estate — you don't have one. And in regulated software the real cost isn't servers, it's **validation labor, implementation time, and operational risk** — which is exactly what we cut: SaaS (no infra to procure or validate), **self-validation automation** (less revalidation effort on every change), **AI-optional resilience** (the deviation→CAPA workflow is the deterministic system-of-record and runs even if the AI is down — Annex 22-correct), and a **90-day bounded sprint** (no multi-year rollout)." **"Does it run if the AI fails?"** > "Yes — AI is advisory with guards and an exception path; the regulated workflow is deterministic and is the system-of-record. Generative AI is *prohibited* in critical GMP decisions under Annex 22 anyway, so a deterministic core isn't a limitation — it's the required design. We turned the rule into resilience." **"Can you run on-prem / air-gapped like they can?"** > "Architecturally yes — we're fully containerized. Today our AI uses cloud providers (Azure OpenAI in-region, data not used for training), which covers privacy for SaaS. A fully air-gapped edition with a self-hosted open model is on our roadmap for buyers who require it." > *Honest guardrail: their self-hosted / air-gapped option is a **genuine advantage** for a large-pharma / data-sovereignty buyer who insists on running it themselves — that is **not** the Lane A buyer (emerging biotech wants SaaS). Concede it gracefully; do **not** claim air-gap parity today.* **Roadmap note:** *Air-gapped / sovereign edition = **Phase 2+** — a **right-sized open model (≈8–70B, not a 756B model)**, self-hosted, added through the existing AI provider gateway as a **governed model-change** (model registry + intended-use + validation + drift/bias). Build it when a customer requires it; not for the Lane A wedge. NB: an open model does not loosen Annex 22 (still advisory-only) and self-hosting increases the validation burden (you own model qualification).* --- ## Where we DON'T fight - Don't claim equal **deployed, reference-proven** maturity — they win on proof. (We match on the QMS *feature set*, not on references or "validated.") - Don't enter a lab/equipment bake-off (LIMS · ELN · QC Planning · eLogbook · CAPS · CPV · CVS) — genuinely out of scope. - Don't pitch rip-and-replace to an entrenched AmpleLogic account (Lane B = overlay, incumbent stays system of record).