One platform, two surfaces. API-side spend visibility for finance and engineering. Consumer-side governance for IT and security. No rip-and-replace.
We're not assuming any of these are true at your company. We're calling 20 teams this month to find out which ones actually are.
"What was last month's $X OpenAI bill actually spent on, by team and use-case?" — and the answer is a spreadsheet.
Your AI bill 3×'d this month and no one can tell you whose experiment did it.
You pay for org-level ChatGPT Team or Claude.ai but you've no idea what data employees send through it.
Your DLP sees the URL but not the prompt; your CASB classifies chat.openai.com as “business productivity.”
Your auditor wants every LLM interaction by [employee] in Q3 — across API and chat — and the answer is “we can't.”
SOC 2, ISO 42001, or the EU AI Act is asking for evidence of AI usage governance and you've nothing to show.
Same identity model. Same classification engine. Same audit ledger. One pane of glass for everyone who pays for AI or worries about it.
Wrap your OpenAI / Anthropic / Bedrock keys — or read from your existing Helicone / Cloudflare AI Gateway / Portkey. Classify every call by SSO group and use-case. Spend dashboard, threshold alerts, hard caps.
Buyer: Head of AI, VP Eng, FinOps lead.
Browser extension or network-layer integration captures employee usage of chat.openai.com, claude.ai, gemini, copilot. Per-user dashboards, category-based policy, audit ledger. SIEM export.
Buyer: CISO, CIO, Head of IT.
Resolv control plane
identity (SSO) ─ classification ─ policy ─ audit ledger
│ │ │
▼ ▼ ▼
Surface A your existing Surface B
API ingest AI infra Consumer governance
(read-mode)
OpenAI Helicone Browser extension
Anthropic CF AI Gateway DNS / CASB hook
Bedrock Portkey Reverse proxy
OpenRouter Langfuse for chat.*
The bills are real. Mid-market and enterprise companies are reporting six- to seven-figure annual AI spend split between API surface and consumer chat. Nobody owns both ends.
The buyer didn't exist a year ago. “Head of AI” and “AI FinOps” are now real budget owners with real questions: where is the spend going, who's driving it, and what data is leaving the building.
Compliance pressure is rising. EU AI Act, ISO 42001, NIST AI RMF, and incoming SOC 2 supplemental controls all require evidence of AI usage governance. “We don't know what employees do on ChatGPT.com” is becoming a board-reportable gap.
Cross-vendor neutrality is the moat. OpenAI ships its own admin tools; Anthropic does the same; Microsoft does the same. The buyer's actual question is “show me everything across vendors” — structurally not any single vendor's product.
The thesis is broad enough that scope-creep is the real risk. Three explicit cuts so we don't pretend otherwise.
LangSmith, Langfuse, Helicone, Adaline, Braintrust own the engineer's prompt-iteration workflow. We sit on top via read-mode, we don't compete.
Nightfall, Microsoft Purview, and Symantec have years of PII regex depth. We integrate them as signal sources rather than reinvent.
Zscaler, Netskope, and Palo Alto Prisma classify by app. We classify by what's in the prompt — one layer down. Complement, not replace.
We're taking 3 design partners for our Surface A beta. Tight loop, direct line to the founder, real influence on what we build.
Form rendered by Tally in light mode for readability.
I'm Bimal Singh. Resolv started as an agent-ops layer for support AI. After ~10 discovery calls, two unprompted signals reframed the problem: nobody owns AI spend visibility end-to-end, and nobody owns shadow-AI governance for consumer chat. Resolv is the platform for both.
If you run AI at a Series B+ tech company or mid-market enterprise — or you're the security / finance person who has to answer for it — I'd like to talk. No deck, no demo (there's nothing to demo yet). Just 30 minutes about how your company actually handles AI cost and risk today.
Book a 30-min call