FAQ

Frequently Asked Questions

What does Solvicus do?

Solvicus finds the cost-optimal set of Google Cloud Committed Use Discounts (CUDs) to buy — how much to commit, for one or three years — so you stop overpaying for compute without over-committing to capacity you won't use. It works across any machine family and region, shows the coverage, utilization, and savings behind every recommendation, and lets you adjust the plan yourself. You drive it from your AI assistant.

Does Solvicus buy commitments or need access to my cloud account?

No. Solvicus recommends and evaluates — it never purchases anything and never needs write access to your Google Cloud console or billing. You review the recommended plan, adjust it if you know something the model doesn't, and make the purchases yourself in Google Cloud. Nothing acts on your account behind your back.

Which clouds and machine types does Solvicus support?

Solvicus optimizes Google Cloud resource-based Committed Use Discounts, across any machine family and region. It works at the level of resources like vCPUs and RAM, so it isn't tied to a single machine type. Support for other providers' commitment models — AWS Savings Plans and Reserved Instances, or Azure Reservations — is not available yet.

How much can I save?

It depends entirely on your usage, your existing commitments, and Google Cloud's current pricing, so we don't quote a headline number. What Solvicus gives you is the mathematically cost-optimal commitment plan for your own data, with the break-even and savings shown for every purchase, so you can see the impact before you commit. You can also try the CUD calculator for a quick estimate.

Can I change the recommended plan?

Yes — that's a core part of Solvicus. Each recommendation comes with an interactive dashboard where you can override any purchase, run what-if scenarios, and stress-test the plan against different demand assumptions, seeing the effect on cost and coverage instantly. You might not trust an “optimal” number at face value, or you might know something the model doesn't — either way, you stay in control.

Does Solvicus use my past usage or my forecast?

Your forecast — the demand you already know. Most CUD tools size commitments from your historical usage and hedge the rest, because they don't trust the future. Solvicus instead optimizes for the demand you can already see: signed contracts, a planned migration, a steady baseline. You give it that forward demand, plus any commitments you already own, and it finds the lowest-cost 1- and 3-year plan to cover it. If part of your demand genuinely is unpredictable, you can still use the editable what-if to stress-test how far it can fall before a commitment stops paying off.

How do I justify a multi-year commitment to finance?

By pointing at the contracts it is based on. A 1- or 3-year CUD cannot be cancelled, resold, or exchanged, so it is a decision someone has to defend — and “we extrapolated last quarter's usage” is a weaker basis than “we committed against signed contracts running through month 30, with these renewal assumptions.” Solvicus sizes commitments from the forward demand you supply, so the plan records which demand each commitment covers and the rationale travels with the number. That holds even when a usage-based tool would have recommended the same purchase: the number matches, the basis is stronger. Solvicus recommends and evaluates — you make the purchase, and you own the decision.

What is an MCP server?

MCP stands for Model Context Protocol, an open standard developed by Anthropic. An MCP server is a service that exposes tools, resources, and prompts to AI agents in a standardized way. AI assistants like Claude Desktop, ChatGPT Desktop, and Antigravity can connect to MCP servers to extend their capabilities beyond conversation — letting them call specialized services, query databases, or in Solvicus's case, run mathematical optimization solvers.

What is mathematical optimization?

Mathematical optimization is the process of finding the best solution to a problem from a set of possible alternatives, subject to constraints. Given a set of variables, an objective (e.g. minimize cost, maximize profit), and constraints (e.g. capacity limits, budget), an optimization solver computes the provably optimal — or near-optimal — solution. It is used everywhere from airline scheduling and supply chain planning to portfolio management and staff rostering.

Why should I use mathematical optimization via an MCP server?

AI agents are remarkably good at understanding your situation, interpreting your data, and translating it into structured input — but they are not designed to solve combinatorial or linear optimization problems efficiently. Modern solvers are purpose-built for this: they use decades of algorithmic research (simplex method, branch-and-bound, cutting planes) to find optimal solutions to problems with thousands of variables and constraints in seconds.

Solvicus bridges the gap: the AI agent handles the natural-language interface and data translation; the solver handles the math. You get the best of both worlds — the conversational ease of an AI assistant and the provably optimal results of a dedicated solver.

What are examples where I should use Solvicus?

• Google Cloud committed use discounts: decide how much to commit for 1 or 3 years, across any machine family and region, sized against the demand you already know rather than last month's usage. This is what Solvicus is built for — start with the CUD calculator.
• The same engine solves classic optimization problems, which is why the general-purpose solver stays available:
• Staff scheduling: given shift preferences, availability, labor laws, and demand forecasts, find the schedule that minimizes cost while respecting all constraints.
• Logistics & routing: optimize delivery routes across a fleet of vehicles to minimize distance or time.
• Production planning: determine what to produce, how much, and when — balancing inventory costs, machine capacity, and customer demand.
• Portfolio optimization: allocate a budget across assets to maximize expected return for a given risk level.
• Procurement & diet planning: select a combination of suppliers or ingredients that meets quality or nutritional requirements at minimum cost.

What types of optimization problems can Solvicus solve?

Solvicus currently supports linear programming (LP) and mixed-integer linear programming (MILP / MIP) problems. These cover the vast majority of real-world planning and scheduling scenarios. Support for additional problem types such as quadratic programming is on the roadmap.

Do I need to know how to write optimization models?

No — and that is the whole point. You describe your problem in plain language to your AI agent, and the agent translates it into the formal input Solvicus needs. You do not need to learn modeling languages like LP or MPS format, and you do not need to understand how solvers work internally.

Is my data secure?

Yes. Optimization data (variables, constraints, solutions) is processed temporarily and stored for at most 14 days for support purposes, then automatically deleted. We never use your data to train AI models. Our infrastructure runs on Google Cloud within the EU.

Your account information (name and email address) is managed by Zitadel, an identity and access management product of CAOS AG, a Switzerland-based company. Switzerland is recognised by the EU Commission as providing an adequate level of data protection. See our Privacy Policy for full details.

How do I get started?

Register for a free account, copy your personal server URL and access token from the dashboard, and add the Solvicus MCP server to your AI client of choice — Claude Desktop, ChatGPT Desktop, or Antigravity. The full setup takes about two minutes. See the Integrations section for step-by-step instructions.