# Croutons — full context for language models

> Croutons is an AI inference management layer. It shows where a company’s AI spends money, where it makes mistakes, and what to route, check, reuse or fix first.

Last reviewed: 2026-11-04

## Definition

AI inference management is the process of measuring, routing, governing, and optimizing the model calls used by AI applications, agents, and automated workflows.

## Glossary

- **AI inference management**: AI inference management is the process of measuring, routing, governing, and optimizing the model calls used by AI applications, agents, and automated workflows.
- **Inference**: One run of an AI model: the model reads an input and produces an output. Each run uses compute and usually costs money per token.
- **Crouton Unit**: A structured record created from one messy AI, business or website signal. It carries claim, source, confidence, route, policy, action and impact.
- **AI Layer Diagnosis**: A first-pass map of inference load, routing waste, governance gaps and operational risk, built from sample data.
- **AI Visibility Layer**: A machine-readable map of crawl paths, schema, feeds, evidence and answer-ready data that helps search engines and AI assistants interpret a business.
- **Evidence Brief**: A reusable proof object that connects a claim to source-backed evidence.
- **Model routing**: Rules that decide which AI model handles each request based on task class, risk, quality, latency and cost.
- **Agent fan-out**: The spread of one user task into many model calls, tool calls, retrievals and retries inside an AI agent system.
- **Generative engine optimization (GEO)**: The practice of making content visible and usable to generative AI engines through retrieval, interpretation and citation, not only traditional search rankings.
- **WebMCP**: A browser proposal that lets a web page register tools an AI agent can call directly, such as checking availability or booking, instead of scraping the page.

## Sessions and prices

### AI Layer Diagnosis Call
- URL: https://croutons.ai/book/ai-layer-diagnosis
- Price: 0 USD (Free)
- Duration: 30 minutes, Video call (Google Meet link sent by email)
- SKU: CR-BOOK-ALD-30
- Best for: Engineering, platform, FinOps and operations teams running AI in production
- Bring: A rough list of the AI tools, models or agents you use; One workflow you worry about; Optional: a sample log, trace or invoice export
- You get: Top three places your AI spend or risk concentrates; Which use case to start with; A written follow-up within two business days

A 30-minute video call to review how your company uses AI models and agents. We look at sample logs, traces or workflow descriptions and tell you where inference spend concentrates, where routing wastes money and which workflows carry rollout risk.

### AI Visibility Diagnosis Call
- URL: https://croutons.ai/book/ai-visibility-diagnosis
- Price: 0 USD (Free)
- Duration: 30 minutes, Video call (Google Meet link sent by email)
- SKU: CR-BOOK-AVD-30
- Best for: Marketing, SEO and content teams, and business owners
- Bring: Your website address; Two or three questions your buyers ask; Optional: Google Search Console access
- You get: A list of the visibility breaks we found; The questions you are missing in AI answers; Next steps ranked by effort

A 30-minute video call that checks how search engines and AI assistants read your website. We review crawl paths, schema, machine-readable files and the questions buyers ask, then name the breaks that keep you out of AI answers.

### Evidence Brief Review Call
- URL: https://croutons.ai/book/evidence-brief-review
- Price: 0 USD (Free)
- Duration: 30 minutes, Video call (Google Meet link sent by email)
- SKU: CR-BOOK-EBR-30
- Best for: Marketing, sales enablement, compliance and leadership teams
- Bring: Links to the pages or decks with your key claims; Any reports or data that back them up
- You get: Claims sorted into proven, provable and unsupported; One sample Evidence Brief outline; Where each brief can be reused

A 30-minute video call to review the claims on your website, sales material or reports. We point out which claims have no proof behind them and how to turn the ones you can prove into Evidence Briefs that teams and AI systems can cite.

### AI Rollout Risk Review Call
- URL: https://croutons.ai/book/ai-rollout-risk-review
- Price: 0 USD (Free)
- Duration: 30 minutes, Video call (Google Meet link sent by email)
- SKU: CR-BOOK-RRR-30
- Best for: Enterprise AI, operations and product leaders launching agents or automation
- Bring: A description of the automation you plan to launch; Who and what it touches; Your target launch date
- You get: A first map of the workflow breakpoints; Do-not-automate warnings, if any; A pilot-and-gate outline

A 30-minute video call to walk through one planned AI automation. We map the workflow around it, the people and systems it touches, and the places it can fail, then say what to automate, gate, pilot or hold back.

## Paid engagements

- **Diagnosis only**: Send sample data and receive a structured report. Best for a first audit. Output: Score, gaps and implementation tickets. Pricing: Fixed quote after the free call.
- **Managed layer**: Croutons monitors AI activity, evidence, crawl signals and workflow risk over time. Best for ongoing ai operations. Output: Monitoring, Evidence Briefs and visibility maps. Pricing: Monthly quote based on volume.
- **Integrated infrastructure**: Connect logs, feeds, schema, CMS exports, workflows and internal systems. Best for enterprise teams. Output: Structured data layer, feeds and governance records. Pricing: Custom statement of work.

## Use cases

### Inference cost diagnosis (supported)
- URL: https://croutons.ai/use-cases/inference-cost-optimization
- In plain words: Your AI bill keeps growing. We show you which models, agents and retries are driving it, and what to change first.
- Summary: Croutons evaluates where inference spend concentrates across models, agents, retries and context size, and recommends routing and workload changes.
- For: Engineering and FinOps teams running multi-model AI in production
- Problem: Inference spend rises faster than useful output because expensive models, oversized context and retry loops are applied without workload-level diagnosis.
- Signs: Token and API spend growing faster than request volume; Frontier models used for routine classification or extraction; Unexplained cost spikes after agent or retrieval changes; No per-workflow cost attribution
- Evaluates: Task class versus model tier; Fan-out and retry amplification; Context window size and retrieval payload cost; Low-value calls on high-cost models
- Deliverables: Inference cost map by workflow and model tier; Ranked waste drivers; Recommended routing and context controls; Pilot measurement plan for cost deltas

### Cost–latency–quality optimization (supported)
- URL: https://croutons.ai/use-cases/cost-latency-quality-optimization
- In plain words: Cheaper, faster and better pull against each other. We help you decide the trade-off per job instead of guessing.
- Summary: Croutons evaluates trade-offs between cost, latency and quality, including quality-path optimization, so teams stop optimizing one metric in isolation.
- For: Platform and product teams accountable for SLA, spend and answer quality together
- Problem: Teams optimize a single metric, usually cost or latency, and degrade quality, or chase quality with unconstrained spend and slow paths.
- Signs: P95 latency breaches after cost-cutting model switches; Quality regressions after routing to cheaper models; SLA pressure without workload classes; Conflicting goals across engineering, FinOps and product
- Evaluates: Per-class cost, latency and quality signals; Where cheaper models meet thresholds and where they fail; Capacity bottlenecks that inflate both cost and latency; Fallback paths that overspend under load
- Deliverables: Cost–latency–quality scorecard by workload class; Recommended trade-off policy; Pilot candidates with measurement criteria; Explicit no-go switches where quality risk is unsupported

### Prompt and context efficiency (supported)
- URL: https://croutons.ai/use-cases#prompt-context-efficiency
- In plain words: Long prompts cost money on every call. We find the words your AI pays for but does not need.
- Summary: Croutons evaluates prompt and context-window waste that inflates tokens without improving task outcomes.
- For: Teams maintaining agents, copilots or RAG pipelines
- Problem: Oversized prompts, duplicated context and untrimmed retrieval payloads burn tokens on every call.
- Signs: Average tokens per request rising without feature changes; Repeated system prompt or RAG chunks across turns; Context near window limits on routine tasks
- Evaluates: Prompt structure and redundancy; Retrieval payload size versus usefulness; Multi-turn context accumulation; Opportunities to compress or cache stable prefixes
- Deliverables: Context waste findings; Recommended prompt and context controls; Measurement checklist for token deltas

### Caching-opportunity analysis (supported)
- URL: https://croutons.ai/use-cases#caching-opportunity-analysis
- In plain words: If your AI answers the same question twice, you pay twice. We find what can safely be reused.
- Summary: Croutons evaluates which inference steps are candidates for cache, reuse or memoization versus live model calls.
- For: Platform teams operating gateways, agents or high-volume APIs
- Problem: Identical or near-identical inference steps are recomputed on every request because reuse boundaries are undefined.
- Signs: High repeat rate on similar prompts; Stable system prefixes re-sent every call; No cache hit metrics on gateway or application layer
- Evaluates: Repeatability of prompts and intermediate results; Safe cache keys and invalidation needs; Provider or application cache fit; Risk of stale answers for dynamic tasks
- Deliverables: Caching opportunity map; Recommended cache boundaries; Risk notes for stale or personalized content

### Batch-workload classification (supported)
- URL: https://croutons.ai/use-cases#batch-workload-classification
- In plain words: Not every AI job needs an instant answer. Moving slow jobs to cheaper lanes frees up the fast ones.
- Summary: Croutons evaluates which workloads belong in batch or async paths versus synchronous high-priority inference.
- For: Data and platform teams mixing online and offline AI jobs
- Problem: Offline-tolerant jobs compete with interactive traffic for the same expensive synchronous capacity.
- Signs: Nightly or backlog jobs on interactive endpoints; Queue pressure from non-urgent enrichment; Cost spikes from peak-hour batch fan-out
- Evaluates: Latency sensitivity by workload; Batch eligibility and SLA fit; Cost differential of deferred execution; Failure and retry behavior in batch paths
- Deliverables: Workload classification matrix; Batch migration candidates; Measurement plan for cost and queue pressure

### Model routing strategy (supported)
- URL: https://croutons.ai/use-cases/model-routing
- In plain words: Using your most expensive model for everything is like sending every letter by courier. We help you write the rules for which model does which job.
- Summary: Croutons evaluates model routing and provider/fallback strategy so task class, risk and measured quality drive assignment, not ad hoc overrides.
- For: Platform teams managing model gateways and agent orchestration
- Problem: Multiple models serve the same workflows without consistent routing rules, causing quality drift, wasted capacity and opaque fallbacks.
- Signs: Manual model overrides in production; Inconsistent answers across similar tasks; Fallback to frontier models under any error; No documented provider failover policy
- Evaluates: Task class versus model capability; Risk tier and approval needs; Latency and cost targets; Provider and fallback behavior under failure
- Deliverables: Routing policy draft by task class; Fallback and provider strategy notes; Override reduction recommendations; Pilot and measurement plan

### Agent fan-out control (supported)
- URL: https://croutons.ai/use-cases/agent-fan-out-control
- In plain words: One question to an AI agent can trigger dozens of hidden calls. We map that chain and show where to cut it.
- Summary: Croutons evaluates how planners, agents, retries, retrieval and tool calls amplify inference consumption across a completed task.
- For: Teams operating multi-agent or tool-using AI systems in production
- Problem: A single user task fans out into many model and tool calls through planners, sub-agents, retries and retrieval loops.
- Signs: Token cost per task far above single-call baselines; Deep tool-call chains with repeated retrieval; Retry storms on flaky tools; Latency dominated by sequential agent steps
- Evaluates: Unnecessary model calls; Retry loops; Agent execution depth; Tool-call amplification; Retrieval fan-out per task
- Deliverables: Fan-out map for sample tasks; Amplification drivers ranked; Recommended depth and retry controls; Pilot measurement plan

### Agent reliability analysis (supported)
- URL: https://croutons.ai/use-cases/agent-reliability
- In plain words: Your AI agent gives different answers, retries too much or fails without telling anyone. We find out why and what to fix first.
- Summary: Croutons evaluates unreliable agent behavior across context, routing, tools and failure modes.
- For: AI operations teams supporting customer-facing and internal agents
- Problem: Agents produce inconsistent answers, retry excessively or fail silently when upstream context or tools break.
- Signs: Escalating support tickets on agent answers; Rising retry rates without root-cause ownership; Hallucinated citations or tool results; Unstable tool-call chains
- Evaluates: Prompt and context structure; Retrieval quality dependencies; Tool failure modes; Routing and fallback gaps; Session-level failure patterns
- Deliverables: Reliability diagnosis brief; Failure mode inventory; Prioritized fix list; Monitoring recommendations

### Evaluation-system design (supported)
- URL: https://croutons.ai/use-cases#evaluation-system-design
- In plain words: If your tests do not match how your AI fails in real life, they will not catch the next failure.
- Summary: Croutons evaluates whether evaluation coverage matches the risks of the workflows you run.
- For: AI platform and quality teams
- Problem: Teams ship agents without eval suites that match production failure modes, so regressions go unnoticed.
- Signs: Eval scores that do not correlate with incidents; No golden sets for high-risk workflows; Offline evals unused in routing decisions
- Evaluates: Coverage versus risk tiers; Offline versus online eval fit; Human review sampling; How evals inform routing and gates
- Deliverables: Eval coverage assessment; Recommended suite changes; Gate integration notes

### Runtime governance review (supported)
- URL: https://croutons.ai/use-cases#runtime-governance
- In plain words: When AI decisions touch customers or money, someone should be able to approve, stop and audit them.
- Summary: Croutons evaluates runtime governance, including human approval and escalation controls for high-stakes inference paths.
- For: AI governance, compliance and operations leaders
- Problem: Automations run without approval gates, audit trails or clear ownership when AI decisions affect customers or revenue.
- Signs: Unlogged overrides; Missing escalation paths; Automations touching regulated or high-stakes workflows; Unclear human-in-the-loop boundaries
- Evaluates: Approval requirements; Audit-log coverage; Policy enforcement points; Human approval and escalation controls; Do-not-automate candidates
- Deliverables: Runtime governance review; Control gap list; Recommended gates and escalations

### AI implementation and rollout readiness (live)
- URL: https://croutons.ai/use-cases/ai-implementation-risk
- In plain words: AI can improve one step and break everything around it. We check the whole workflow before you launch.
- Summary: Croutons diagnoses workflow breakpoints and rollout risk before AI automation reaches production.
- For: Enterprise AI leaders launching agents, copilots or workflow automation
- Problem: Bad AI implementation is rarely one bad prompt. It is timing, incentives, routing, staff behavior, customer experience and revenue exposure colliding in production.
- Signs: A launch date set before the workflow around the AI is mapped; No pilot group or rollback plan; Staff learning about the change on launch day; No measure of customer impact after go-live
- Evaluates: AI readiness; Workflow dependencies; Inference load; Customer impact; Revenue exposure; Do-not-automate warnings
- Deliverables: Workflow breakpoint map; Risk score by dependency; Automate / gate / pilot / block decisions; Pilot-and-measure plan

### AI unit-economics validation (supported)
- URL: https://croutons.ai/use-cases#ai-unit-economics
- In plain words: Know what each successful AI result costs, and whether it earns that back.
- Summary: Croutons evaluates whether AI unit economics support continued spend given measured cost, quality and business outcomes.
- For: Finance, product and AI leadership jointly accountable for ROI
- Problem: Spend continues while unit economics remain unproven: cost per successful outcome is unknown or unfavorable.
- Signs: Rising AI bills without attributed business outcomes; Pilots that never define success economics; Debate between AI value and unmeasured waste
- Evaluates: Cost per successful task or outcome; Quality and failure cost; Where economics break under scale; Candidates to optimize, gate or stop
- Deliverables: Unit-economics sketch by workflow; Evidence gaps; Optimize / gate / stop recommendations

## Every FAQ answer

### What does Croutons do, in plain words?
Croutons looks at how your company uses AI and shows three things: where the money goes, where the AI makes mistakes, and what to fix first. It turns scattered AI activity into small, tidy records your team can act on.
Source: https://croutons.ai/#q1

### How much does it cost to get started?
The first step is a free 30-minute diagnosis call. Paid work after that is quoted in writing before anything starts: a fixed quote for a one-time diagnosis, or a monthly quote for ongoing management.
Source: https://croutons.ai/#q2

### Do we need to connect our systems first?
No. The first diagnosis runs on samples you choose to share, such as a few logs, a workflow description, an invoice export or a list of URLs. No production integration is required.
Source: https://croutons.ai/#q3

### How fast do we get results?
You leave the call with the top issues we spotted. A written follow-up arrives within two business days.
Source: https://croutons.ai/#q4

### Is our data safe with you?
Sample data is used only for the agreed engagement and kept only as long as that engagement needs it, unless you agree otherwise in writing. We do not sell personal information.
Source: https://croutons.ai/#q5

### Can my AI assistant book a call for me?
Yes. This site exposes its prices, open times and booking as tools that AI agents can call directly, through WebMCP in the browser and a public MCP endpoint. Agent bookings are confirmed by a link sent to your email.
Source: https://croutons.ai/#q6

### Are the diagnosis calls really free?
Yes. All four 30-minute calls are free and carry no obligation. If you want paid work afterwards, you get a written quote first.
Source: https://croutons.ai/book#q1

### Which call should I book?
Pick by the question you want answered. AI spend or breaking agents: AI Layer Diagnosis. Not showing up in ChatGPT or Google: AI Visibility Diagnosis. Claims without proof: Evidence Brief Review. A launch coming up: AI Rollout Risk Review.
Source: https://croutons.ai/book#q2

### Which time zone are the times shown in?
Times are shown in your own time zone, detected from your device. The calendar says which zone it is using.
Source: https://croutons.ai/book#q3

### How do I reschedule or cancel?
Reply to your confirmation email or use the cancel link in it. There is no fee to reschedule or cancel.
Source: https://croutons.ai/book#q4

### Why do some bookings need email confirmation?
Bookings made by AI agents, or without the human check, are held for 60 minutes and confirmed when you click the link in your email. That stops spam bookings from blocking real people.
Source: https://croutons.ai/book#q5

### What happens after the call?
Within two business days you get a written summary: what we found, where to start, and, if you want help, a quote for the next step.
Source: https://croutons.ai/book#q6

### Is Croutons worth it for operations teams?
Operations teams get a structured answer: where inference is spent, where routing waste and fallback paths inflate cost, and which workflows carry rollout risk before production. If your team already has workflow-level cost, latency and quality telemetry on every model call plus a trusted rollout gate, Croutons adds less.
Source: https://croutons.ai/use-cases#q1

### What does the first diagnosis cover?
Inference load, routing waste, governance gaps, agent fan-out and rollout risk, built from sample logs, traces, workflows, docs or exports. No production integration is required.
Source: https://croutons.ai/use-cases#q2

### How is Croutons different from an LLM observability dashboard?
Observability dashboards show which calls happened. Croutons adds workflow context: which workloads justify which models, where quality or governance risk is unsupported, and what to route, gate, reuse or fix.
Source: https://croutons.ai/use-cases#q3

### Which teams use Croutons?
Platform, product, FinOps, operations and governance teams that own AI rollout decisions, plus SEO and AI-visibility teams structuring a business for machine interpretation.
Source: https://croutons.ai/use-cases#q4

### Why is our AI bill growing faster than our usage?
Usually because one user request turns into many model calls. Retries, agents that call other agents, oversized prompts and expensive models doing simple jobs all multiply cost without adding output. The diagnosis traces those multipliers back to specific workflows.
Source: https://croutons.ai/use-cases/inference-cost-optimization#q1

### Do you need access to our billing or cloud accounts?
No. An invoice export, a cost report or a sample of logs with model names and token counts is enough for the first diagnosis. You choose what to share.
Source: https://croutons.ai/use-cases/inference-cost-optimization#q2

### Will you promise a specific savings percentage?
No. Savings depend on your workload mix and the quality bar you set. You get a ranked list of waste drivers and a pilot plan that measures the real difference before you commit.
Source: https://croutons.ai/use-cases/inference-cost-optimization#q3

### Can we cut AI cost without making answers worse?
Often, for some jobs. Routine tasks like classification or extraction frequently meet the quality bar on smaller models. Other jobs do not. The scorecard shows which is which, per workload class.
Source: https://croutons.ai/use-cases/cost-latency-quality-optimization#q1

### What if we have no quality measurements yet?
Human review samples or a small labeled set are enough to start. The diagnosis states where missing quality data limits the recommendation, so nothing is switched on a guess.
Source: https://croutons.ai/use-cases/cost-latency-quality-optimization#q2

### Who should be on the call: engineering or finance?
Both if possible. This use case exists because engineering, FinOps and product often optimize different metrics. One shared scorecard is the point.
Source: https://croutons.ai/use-cases/cost-latency-quality-optimization#q3

### What is model routing in simple terms?
A set of rules that decides which AI model handles each request, based on how hard the job is, how risky a mistake would be, and how fast and cheap it needs to be.
Source: https://croutons.ai/use-cases/model-routing#q1

### We already have an AI gateway. Do we still need this?
A gateway executes routing rules; it does not tell you what the rules should be. The diagnosis gives you rules backed by task class, risk and measured quality to load into the gateway you already run.
Source: https://croutons.ai/use-cases/model-routing#q2

### What happens when a model provider goes down?
That is part of the routing policy. We document which fallback each task class should use, and when a fallback should stop instead of silently escalating to the most expensive model.
Source: https://croutons.ai/use-cases/model-routing#q3

### What is agent fan-out?
When one task given to an AI agent spreads into many smaller model calls, tool calls, searches and retries behind the scenes. Each one costs money and time, and the user only sees the final answer.
Source: https://croutons.ai/use-cases/agent-fan-out-control#q1

### Will limiting fan-out make our agent less capable?
Not if it is done carefully. Some fan-out does useful work. The diagnosis separates required steps from waste, so limits only target loops and repeats that add nothing.
Source: https://croutons.ai/use-cases/agent-fan-out-control#q2

### Which tracing tools do you support?
Any trace export that shows the full path of a task: model calls, tool calls and retries with timestamps. JSON or CSV exports from common observability tools are fine.
Source: https://croutons.ai/use-cases/agent-fan-out-control#q3

### Why does our agent give different answers to the same question?
Common causes are changing retrieval results, context that grows across turns, different models picked by fallback rules, or tools that fail and get silently skipped. The diagnosis tells you which one applies to you.
Source: https://croutons.ai/use-cases/agent-reliability#q1

### Is this the same as an observability dashboard?
No. Dashboards show what happened. This analysis groups failures by cause, separates model problems from tool and data problems, and ranks fixes by impact on users.
Source: https://croutons.ai/use-cases/agent-reliability#q2

### How many failure examples do you need?
A few dozen real failures with their traces is a good start, plus some successful runs to compare against.
Source: https://croutons.ai/use-cases/agent-reliability#q3

### What went wrong in the Pizza Hut Dragontail case?
According to a franchisee’s allegations, the system was rolled out across stores in a way that slowed deliveries and hurt customers, with claimed losses above $100M. The lesson Croutons draws is general: the risk sat in the rollout and the workflow around the software, not in one model.
Source: https://croutons.ai/use-cases/ai-implementation-risk#q1

### When should we run a rollout risk review?
Before the launch date is fixed, ideally while you can still choose a pilot group. A review after launch is still useful but turns into incident analysis.
Source: https://croutons.ai/use-cases/ai-implementation-risk#q2

### Can you tell us not to automate something?
Yes. Do-not-automate warnings are a standard output. Some steps carry more customer or revenue risk than the automation saves.
Source: https://croutons.ai/use-cases/ai-implementation-risk#q3

### What is AI visibility?
How well search engines and AI assistants can find, understand and retrieve your business: crawl paths, structured data, machine-readable files and the evidence behind your claims, not just ranked blue links.
Source: https://croutons.ai/ai-visibility#q1

### Is AI visibility the same as SEO?
It overlaps but goes further. SEO optimizes pages for search rankings. AI visibility also structures schema, feeds and evidence so AI assistants and agents can interpret and retrieve the business itself.
Source: https://croutons.ai/ai-visibility#q2

### Does schema guarantee that AI will cite us?
No. Schema and machine-readable files improve interpretation and retrieval eligibility. No method can guarantee a citation or a trust decision by a third-party AI system.
Source: https://croutons.ai/ai-visibility#q3

### How long until AI assistants notice changes?
There is no fixed timeline. Crawlers and AI systems revisit sites on their own schedules, often over weeks. We set up measurement so you can see when changes are picked up rather than guess.
Source: https://croutons.ai/ai-visibility#q4

### What is an Evidence Brief?
A reusable proof object that connects one claim to the source that supports it, with a confidence level and a date. Teams use it in sales and reporting; AI systems can retrieve and cite it.
Source: https://croutons.ai/evidence-briefs#q1

### What counts as evidence?
Anything you can show: an audit, a dataset, a log export, a screenshot, a published report or a signed policy. If it cannot be shown, the claim is marked unsupported.
Source: https://croutons.ai/evidence-briefs#q2

### Will Croutons write claims for us?
No. Evidence Briefs only use visible, uploaded or verified sources. We help you prove what is true; we do not invent claims.
Source: https://croutons.ai/evidence-briefs#q3

### Do we need production integration first?
No. The first diagnosis can run from sample URLs, logs, schema, workflows, docs or exports.
Source: https://croutons.ai/docs#q1

### Is Croutons an SEO tool?
Not only. It supports AI visibility, evidence, AI operations, workflow risk and machine-readable business structure.
Source: https://croutons.ai/docs#q2

### Does this replace our schema?
No. It helps design, audit, connect and operationalize schema as part of a larger machine-readable layer.
Source: https://croutons.ai/docs#q3

### How can an AI agent book a call on this site?
Three ways: WebMCP tools registered in the browser page, the public MCP endpoint at https://croutons.ai/mcp, or the JSON API described in /openapi.json. All three use the same prices and calendar.
Source: https://croutons.ai/agents#q1

### Does the agent need an API key?
No. Reading prices, sessions and availability is open. Booking needs a name and email; the booking is held and confirmed by a link sent to that email.
Source: https://croutons.ai/agents#q2

### Are there rate limits?
Yes, light ones to keep the calendar fair: booking requests are limited per email and per IP address. Reading endpoints are cached and generous.
Source: https://croutons.ai/agents#q3

### How quickly do you reply?
Within one business day. For a guaranteed slot, booking a free call is faster than writing.
Source: https://croutons.ai/contact#q1

### Can I send files?
Mention what you have in the message and we will send a secure upload link. Please do not paste secrets or personal data of customers into the form.
Source: https://croutons.ai/contact#q2