SpendLil watches every AI request across your business — the spend, the risk, the data. Add one header. That's it. Your automated AI officer, always available, never blocking your AI.
HIGH risk systems must comply or face fines up to €30M or 6% of global turnover.
Am I affected?Works with every major AI provider
The reality, right now
Your team adopted AI quietly, at the keyboard level. The spend is scattered. The data risk is invisible. Nobody has the full picture — because the oversight layer was never built.
of UK SMBs have no visibility into AI spend
average hidden monthly AI cost per company
of AI interactions contain some form of PII
API keys scattered across teams, personal subscriptions, ad-hoc tools — AI spend is invisible until the invoices arrive.
Customer names, NI numbers, even credit card details get pasted into AI prompts daily by well-meaning staff who don't know the risk.
Staff sign up for GPT-4, Claude, Gemini without telling IT. New models appear every week — your team adopts them before you've assessed them.
Features
One platform. Full visibility. No surprises.
Total spend, per-key breakdown, model costs, provider comparison, and forecasts. Know your daily burn rate at a glance.
Pass X-SpendLil-Tag on any request to label it — chatbot, support, sales, customer name. Filter spend by tag in the dashboard.
Scans every request and response for emails, NI numbers, credit cards, postcodes. Alerts fire immediately on detection.
Set daily or monthly budgets per account. Get alerted when you approach limits — before the surprise invoice arrives.
Budget thresholds, new keys, cost spikes — via email or Slack on Growth and above.
Generate PDF reports covering spend, PII exposure, and compliance status.
Track EU AI Act readiness across all key articles. Build your audit trail automatically.
New API key used for the first time? SpendLil creates the record automatically. Zero setup.
Fire-and-forget logging — usage is written after your response returns, never blocking it.
If SpendLil goes down, your AI keeps running. The key is in your request — not stored with us.
How It Works
No agents to create. No keys to manage. No setup wizard. Add one header and you're tracking.
Register in 30 seconds. You'll get an account key — sl_abc123. That's the only credential you manage.
Point your existing AI SDK at your provider's SpendLil gateway — e.g. openai.gateway.spendlil.ai instead of api.openai.com. Add X-SpendLil-Key. Done.
Your first request auto-discovers the API key. Spend, model costs, and tags appear in real-time. No setup, no agents to create.
from openai import OpenAI
client = OpenAI(
api_key="sk-your-openai-key",
base_url="https://openai.gateway.spendlil.ai/v1",
default_headers={
"X-SpendLil-Key": "sl_abc123",
"X-SpendLil-Tag": "my-chatbot", # optional
},
)
resp = client.chat.completions.create(
model="gpt-4o",
messages=[{"role": "user", "content": "Hello!"}],
)
# Response identical to calling OpenAI directly
# + header: X-SpendLil-Route: governedPricing
Start free. Upgrade when you need more. No credit card required.
Get started with AI visibility
For teams getting serious about AI
For SMBs that need full governance
For regulated industries
Private Cloud & Self-Hosted
For regulated industries — finance, legal, healthcare, government — where data sovereignty is non-negotiable. Full stack deployment. White-labelling available. Annual licence pricing.
August 2026 is closer than it looks. SpendLil automates EU AI Act tracking, risk registers, and audit trails — so you're covered without the headcount.
All staff using AI must have documented training. SpendLil tracks training records per employee.
High-risk systems need a risk register with likelihood, impact, and mitigations documented.
Training data must be documented, including sources, bias assessments, and personal data handling.
Users must be told when they're interacting with AI. Disclosure notices tracked per key.
High-risk decisions need human review. SpendLil tracks approval workflows and escalation logs.
Organisations using high-risk AI must maintain logs and cooperation evidence for authorities.
High-risk AI systems must be registered in the EU database. SpendLil generates the export.
Serious incidents must be reported within 72 hours. Full incident log built in from day one.
SpendLil evaluates your current setup against all key EU AI Act requirements and shows exactly what gaps to close before August 2026.
FAQ
Point your existing AI SDK at your provider's SpendLil gateway — e.g. openai.gateway.spendlil.ai instead of api.openai.com. Add one header: X-SpendLil-Key. Every request is proxied, logged, and costed. The response you get back is identical to calling the provider directly.
Never. Your provider API key passes through SpendLil with every request and is never stored. We store a SHA-256 hash and the last 4 characters for identification only. This is a core design principle, not an afterthought.
Your AI keeps running. Because your API key is in every request, you can always fall back to calling your provider directly. Even a full outage means you lose spend tracking temporarily, not your AI functionality.
Yes — OpenAI, Anthropic, Google, Mistral, Cohere, and any OpenAI-compatible API. If it uses the standard chat completions format, we support it.
An optional header you can pass on any request to label it — "chatbot", "support", "customer-123", "batch-job". Tags appear in the usage explorer and let you filter spend by your own identifiers.
Minimal. SpendLil runs on Lambda in eu-west-2 (London). Typical added latency is 20-50ms. Logging is fire-and-forget — usage records are written after your response returns, never blocking your AI request.
We scan request and response bodies for UK-specific PII patterns — NI numbers, postcodes, UK phone numbers, credit card numbers, and email addresses. High-confidence matches alert immediately.
Yes — use X-SpendLil-Tag to separate environments (production, staging, dev) or teams (support, sales, engineering). Filter and break down spend by tag in the dashboard.
SpendLil is hosted in AWS eu-west-2 (London). We store usage metadata — token counts, costs, model names, timestamps — never full conversation content. Data retention is configurable.
Yes — we offer private cloud deployments into your own AWS, Azure, or GCP account. Your data never leaves your environment. Annual licence pricing. Contact hello@spendlil.ai.
Your team is already using AI. SpendLil makes sure you're in the room when they do — watching the spend, the risk, and the data.
Free plan available · No credit card needed · 2-minute setup