# Sp · Spend Management — element 50 of 58 > Stop the subscription bleed. Turns AI sprawl into a managed budget. - **Group:** 11 · Trust & Compliance - **Necessity:** Optional - **Price band:** $ · under $30/mo - **Maturity:** Experimental - **Edition:** v2026.Q3 · verified 2026-09-13 ## Leading tools (v2026.Q3) - **Ramp** — control at the card - **Torii** — the dedicated smp pick - **Vertice (incl. Vendr)** — negotiation as a service - **Vantage** — watch the token bill - **Cledara** — card-first saas control ## Our take The average portfolio carries seven redundant AI tools. A quarterly cull funded by this element pays your entire stack. ## Combines with Ac, Ts ## The top 5 — deep dossier (verified 2026-09-13) Ramp, plus a spreadsheet — that's the honest answer for an AI-led startup in 2026. Under ~25 people, a spreadsheet and one virtual card per vendor IS your spend management; Ramp's $0 tier gives you the cards, the bill pay, and — since mid-2026 — AI token spend tracking, which is where the bleed actually is now (78% of IT leaders got surprise AI charges per Zylo's 2026 Index). Torii is the pick when you cross ~100 employees and need a dedicated SaaS+AI management platform; Vertice (which now owns Vendr) when annual SaaS spend passes ~$500k and negotiation beats cancellation; Vantage when the bleed is API tokens, not seats; Cledara when you want card-level enforcement without US-fintech lock-in. The enterprise procurement suites (Zip, Tropic, Zylo) are where you'll end up, not where you start. 1. **Ramp** (Ramp) — Free $0 · Plus $15/user/mo + platform fee · Enterprise custom · Procurement as add-on. Best for: Startups that want spend control as a side effect of how they already pay — unlimited virtual cards, bill pay, and AI-agent expense review at $0. Why: The default because control lives at the payment layer: one card per vendor means nothing renews without you seeing it, and the free tier includes automated invoice processing and AI expense review. In 2026 Ramp leaned directly into this element's new problem — AI token spend management tools (CEO on CNBC, Jul 2026) — backed by transaction data across 70,000+ customers and a $44B valuation (Jun 2026). Watch: It's a finance platform first — no usage/SSO-level discovery, so shadow AI signed up with a personal card or free tier stays invisible. Procurement is a paid add-on, and the platform-fee math on Plus is opaque until you talk to sales. [https://ramp.com](https://ramp.com) 2. **Torii** (Torii Labs) — Custom quote, priced to environment · three products: SaaS Management, Identity Governance, AI Management · no free tier. Best for: 100+-employee companies that need real discovery — SSO, finance, browser extension — across SaaS and AI apps, plus token-level AI spend visibility. Why: The canonical dedicated SMP for this element: a Leader in the July 2026 Gartner Magic Quadrant for SaaS Management Platforms, and the first of the class to ship a real AI Management product — AI app/project discovery, token consumption tracking, and spend by department, with a dashboard showing 'which users are burning through tokens and what models are being overused.' Claims 25% cost reduction via continuous discovery and license reclamation. Watch: Quote-only pricing with no self-serve tier — overkill below ~100 employees. Like every SMP it's consolidating toward identity governance, so you may be buying more platform than you need for pure spend control. [https://www.toriihq.com](https://www.toriihq.com) 3. **Vertice (incl. Vendr)** (Vertice) — Custom, savings-backed ('hard cost savings contractually guaranteed') · Vendr free price checks · SaaS purchasing + cloud cost optimization. Best for: Companies past ~$500k annual SaaS spend where the win is negotiating renewals, not canceling seats. Why: Negotiation-as-a-service with the category's best data moat after acquiring Vendr: pricing benchmarks across 32,000+ vendors on the Vertice side plus Vendr's 130,000+ real negotiations across 4,500+ suppliers. Vertice claims 20% average savings, 7x ROI, and 90-day payback, contractually guaranteed; Vendr's 5,000+ teams averaged 18.4% savings per transaction — numbers that comfortably clear the fee at real spend levels. Watch: Both savings figures are first-party marketing, computed against list prices you might never have paid. The Vendr integration is fresh — announced without public deal terms or a clear product roadmap — and quote-only pricing makes the ROI claim hard to pre-verify. [https://www.vertice.one](https://www.vertice.one) 4. **Vantage** (Vantage) — Starter free (≤$2.5k tracked spend) · Pro $30/mo (≤$7.5k) · Business $200/mo (≤$20k) · Enterprise custom. Best for: AI-native startups whose real bleed is API tokens and cloud, not seats — one view across OpenAI, Anthropic, Cursor, and 30+ providers. Why: The subscription-management tools mostly can't see your biggest AI line item: metered API spend. Vantage ingests 30+ providers — explicitly including OpenAI, Anthropic, and Cursor — next to AWS/GCP/Azure, with anomaly alerts, an MCP server, and an automated FinOps agent on Enterprise. The free tier means a seed-stage team gets provider-level token cost visibility for $0. Watch: It's FinOps, not SaaS management — no license reclamation, no renewal calendar, no shadow-IT discovery. Tracked-spend tiers escalate fast once cloud spend is real, and self-serve tiers cap data retention at 6–12 months. [https://www.vantage.sh](https://www.vantage.sh) 5. **Cledara** (Cledara) — Basic £100/mo (free if 10+ apps paid via Cledara cards in month one) · Premium/Pro custom · add-ons £150–200/mo · 16% off annual. Best for: 25–150-person companies (especially UK/EU) that want Ramp-style card-level enforcement purpose-built for software subscriptions. Why: The cleanest 'stop the bleed' mechanic in the category: every subscription gets its own virtual card with limits, approvals, and renewal alerts, so cancellation is enforcement rather than a support-ticket plea. 1,200+ companies, 700,000+ SaaS transactions analyzed, and a claimed 23% SaaS cost reduction — at a flat entry price a Series A can actually budget. Watch: Card-based discovery misses anything not paid through Cledara, and there's no AI/token spend tracking. Pricing is GBP-denominated and add-on modules (spend optimization, IT, compliance) stack up fast; US-heavy stacks usually default to Ramp instead. [https://www.cledara.com](https://www.cledara.com) ### How to choose - If You're under ~25 people → A spreadsheet plus Ramp's free tier, one virtual card per vendor. A quarterly cull meeting is the whole practice — Zylo's 2026 data shows license utilization is only 54% even at managed enterprises, so assume a third of your seats are dead. - If Your AI bill is metered API tokens (OpenAI, Anthropic, inference), not seats → Vantage free tier from day one — 78% of IT leaders hit unexpected AI/consumption charges in 2025-26 (Zylo 2026 Index), and card-level tools only see the invoice after it lands. - If You're 25–150 people and subscriptions multiply faster than you can track → Card-level enforcement: Ramp (US-default) or Cledara (£100/mo, subscription-native, UK/EU) — cancellation you enforce beats cancellation you request. - If Annual SaaS spend passed ~$500k and renewals dominate → Add a negotiation layer — Vertice's contractually guaranteed savings or Vendr's free price checks; 18–20% average claimed savings clears the fee at that spend level, but benchmark the benchmark. - If You're 100+ employees with real shadow-IT/shadow-AI risk → A dedicated SMP — Torii first look (MQ Leader, token-level AI tracking), CloudEagle or BetterCloud as bake-off candidates; expect quote-based pricing everywhere. ### The field (19 more) Zip, Tropic, Zylo, CloudZero, Spendflo, CloudEagle.ai, BetterCloud (acquired), Brex, NachoNacho, Josys, Lumos, Productiv (fading), Vendr (acquired), Sastrify (acquired), Trelica (acquired), Airbase (acquired), Intello (acquired), Blissfully (dead), G2 Track (dead) Full dossier data: https://elems.ai/e/sp.json --- Source: [elems.ai](https://elems.ai/e/sp.html) — the periodic table of the AI-led startup. Data: https://elems.ai/elements.json (CC BY 4.0, cite elems.ai).