> For the complete documentation index, see [llms.txt](https://docs.archera.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.archera.ai/help-center/user-guide/archera-ai-prompt-guide.md). # Archera AI Prompt Guide A library of copy-pasteable prompts for Archera AI, the agent for cloud rate optimization - organized by what you are trying to get done Archera AI is Archera's agent for cloud rate optimization. It reads your real account data through the same APIs the app uses, so it can answer questions about your spend, your commitments, and your savings — and, when you ask it to, build and apply commitment plans. This guide is a library of copy-pasteable prompts, organized by what you are trying to get done. Every prompt works on any surface where Archera AI runs. ## Where Archera AI Runs | Surface | How you reach it | Notes | | ------------------- | ---------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | In the Archera app | The Archera AI panel (dock it beside the page, or open the full-page view) | The only surface that can navigate the app for you, open modals, invite a teammate, or hand you to live support. It also knows which page you are on, so "explain this plan" works with no IDs. | | Slack | The Archera assistant pane, or `@Archera` in any channel | The thread is the memory — each reply re-reads the thread. Use `/archera-org` to set which Archera organization a channel uses. Works in Slack Connect (shared) channels: you are answered as your own org. | | Your own MCP client | Connect Archera's MCP server (Claude Code, Claude Desktop, any MCP-capable client) | Same data tools, no app-navigation tools. Good for pulling Archera numbers into work you are already doing elsewhere. | Archera AI always runs as you: your organization, your role, your permissions. If your role cannot see invoices, neither can Archera AI when it is answering you. To set up the MCP option, see [Archera MCP Server — Connect in Claude](/help-center/faq/mcp-server.md). ## How to Write a Good Prompt A strong prompt names three things. You do not need all three every time, but adding the missing one is almost always what turns a vague answer into a useful one. * **Intent** — the question or the action. "Which commitments are underutilized?", "Build a renewal plan." * **Scope** — provider, date range, account, segment, service, or a specific plan or commitment. "AWS only, last 45 days, production segment." * **Direction** — the shape of the answer. "As a table sorted by savings.", "Two sentences, no jargon.", "CSV I can paste into a sheet." Intent + Scope + Direction: ``` Show me uncovered reservable spend (intent) for AWS over the last 30 days, grouped by service (scope), as a table sorted by cost descending (direction). ``` ### Tips * Ask for the reasoning, not just the number. "…and show me which fields you used" keeps the answer auditable. * Say "verify that" to force a fresh tool call instead of a restatement of an earlier answer. * Paste provider IDs freely. If you paste an AWS Organizations ID or a `provider_reservation_id`, Archera AI resolves it to the matching Archera record itself. * Attach files. Images, PDFs, CSV/TSV, JSON, Markdown, and plain text are read directly. Per-file limits: 5 MB images, 10 MB PDFs (50 pages), 2 MB text. * Chain follow-ups. Archera AI keeps the thread's context; "now compare that to 3-year no-upfront" is a complete prompt after a plan question. ## 1. Get Oriented Start here when you do not know what to ask yet. `overview` returns a provider-level snapshot — savings rate, coverage, utilization, missed savings, and prioritized next actions. Tools: `overview`, `list_organizations`, `list_integrations`, `get_integration` ``` How am I doing on AWS? Give me the headline numbers and the top three things you would fix first. ``` ``` Summarize my Azure position: coverage, utilization, and any missed savings. Then tell me what data you do not have. ``` ``` Which cloud accounts and integrations are connected, and since when has cost data been available for each? ``` ``` I am new here. Explain in plain language what Archera is doing for my account right now, and what I would have to decide next. ``` ## 2. Explore Cloud Costs Two different lenses, and picking the right one matters: * Total spend, including non-reservable cost (data transfer, support, tax, marketplace) → `cost_explorer`. * Reservable spend only, split by whether a commitment already covers it → `daily_reservable_cost`. This is the spend Archera can act on. Tools: `cost_explorer`, `cost_explorer_filters`, `daily_reservable_cost`, `daily_coverage` ``` What are my biggest cost drivers on AWS over the last 30 days? Group by service. ``` ``` How has my total cloud spend changed month over month for the last six months? Call out how much of the change was reservable versus non-reservable. ``` ``` Show me daily reservable cost for AWS over the last 30 days, split by uncovered versus commitment-covered, grouped by service. ``` ``` Which services still have on-demand reservable cost? Sort by monthly amount descending and tell me which are worth committing to. ``` ``` What filters can I use in Cost Explorer for Azure? Then apply them to show me spend for just my production subscriptions. ``` > **Note** — Cloud providers publish cost data with roughly a 1.5-day lag, and Archera AI's date ranges use an exclusive end date. If you ask "through today," expect the last day or two to be missing; Archera AI will tell you the effective range it actually used. ## 3. Review Your Commitment Portfolio Tools: `commitments_summary`, `commitments`, `commitment_details`, `commitment_attributes`, `commitments_metrics`, `daily_coverage`, `commitment_types` ``` Give me a portfolio summary of my existing commitments: how many, what terms, total monthly commitment cost, and blended utilization. ``` ``` Do I have any underutilized commitments? Show utilization per commitment, worst first, and flag anything under 90%. ``` ``` Which commitments expire in the next 90 days? Include term, monthly cost, and what they are covering. ``` ``` Show commitment savings and utilization for the last 45 days with daily detail. Are we trending up or down? ``` ``` Split my savings last month into insured versus uninsured. Explain which commitments fall into each bucket. ``` ``` Explain this commitment: what is it covering, when does the lock-in end, and how much has it saved so far? ``` > **Tip** — "Insured" means Archera Guaranteed Commitments (GRIs, GSPs, GCUDs) with the moneyback guarantee. "Uninsured" means native cloud commitments (RIs, Savings Plans, CUDs). Archera AI reads this from the data, not from assumptions. ## 4. Plan New Commitments Archera's planner accounts for commitments you already hold and only proposes coverage for uncovered usage. There are three default plans — Recommended (30-day), Balanced (1-year), and High Savings (3-year). Tools: `commitment_plan_recommended`, `commitment_plans_default`, `list_commitment_plans`, `commitment_plan_details`, `commitment_plan_line_items`, `commitment_plan_covered_resources`, `commitment_plan_covered_resources_summary`, `create_commitment_plan` ``` How much would the Recommended Plan save me? Show monthly net savings, term, and any upfront cost. ``` ``` Compare the three default plans for AWS side by side: monthly commitment, net savings, lock-in, and what each one covers. ``` ``` What is being covered in the most flexible plan? List the resources and the coverage each line item is matched to. ``` ``` Walk me through the Recommended Plan line by line. For each line item, tell me the instrument, the quantity, and why the planner chose it. ``` ``` I am considering committing to RDS. Is RDS reservable for my provider, and how much uncovered RDS spend do I have? ``` ``` Build me a plan scoped to my production segment only, 1-year terms, no upfront. ``` > **Note** — Plan line items do not always mirror the resource you asked about one-for-one. That is intentional: the planner sizes to uncovered usage after existing coverage and instance-size flexibility. Ask "why do these line items look different from what I expected?" and Archera AI will pull the covered-resource detail. ## 5. Run What-If Comparisons This is the highest-value category and the easiest one to phrase badly. A term-length question ("what about 3-year?") is not a request to look at a different default plan — the defaults differ in coverage and payment mix too. Say the term and the payment option you want. Tools: `commitment_plan_comparison`, `commitment_comparison`, `catalog_sku_offer_comparison`, `search_catalog_skus`, `catalog_sku_details` ``` Take my Recommended Plan and compare it against 3-year no-upfront. Show plan-wide totals and the delta versus current. ``` ``` For this plan, compare 1-year and 3-year at both no-upfront and all-upfront. Which lands best on net monthly savings, and how many line items would fall back to a shorter term? ``` ``` Just for this one line item: what would happen if I moved it to a 1-year GRI? ``` ``` What is the monthly AWS on-demand cost for 10 m5.large, and what would each commitment option cost instead? ``` ``` Compare AWS commitment options for an m5.large in us-east-1 across all terms and payment options. ``` ``` What is the monthly Azure cost of a Dsv6 Postgres database, and is it reservable? ``` > **Tip** — Comparisons are read-only. Archera AI only creates a saved plan if you ask for one you can keep and apply. ## 6. Edit, Copy, and Apply Plans The default plans are read-only. To change one, ask Archera AI to copy it with your edits — that is a single atomic step. Tools: `copy_commitment_plan`, `update_commitment_plan`, `update_commitment_plan_line_items`, `delete_commitment_plan`, `apply_commitment_plan`, `create_commitment_plan_template`, `list_commitment_plan_templates`, `commitment_plan_template_details`, `update_commitment_plan_template`, `delete_commitment_plan_template` ``` Copy the Recommended Plan and drop the ElastiCache line items. Name it "Q3 compute only". ``` ``` In my saved plan, change every 30-day line item to 1-year and show me the new totals before you save anything. ``` ``` Rename this plan and update its description to say it covers only the production segment. ``` ``` I am ready to buy. Walk me through exactly what applying this plan will do, then wait for my confirmation. ``` ``` Show me my commitment plan templates and what each one automates. ``` > **Note (write actions)** — `apply_commitment_plan`, plan updates and deletes, and all template writes require an explicit confirmation. Archera AI will summarize the change — plan name, monthly commitment, net savings, term, upfront cost — and wait for an unambiguous "yes." It never applies a plan on an ambiguous signal. ## 7. Renewals, Expirations, and Exchanges Tools: `commitment_plan_renewal_commitments`, `create_renewal_commitment_plan`, `exchange_recommendations`, `apply_exchange_recommendations` ``` Which commitments are up for renewal, and what would a renewal plan look like? ``` ``` Build a renewal plan for everything expiring in the next 60 days. Compare it to letting those commitments lapse. ``` ``` Are there Azure reservations I should exchange? Show the recommendation, the savings delta, and what I would be giving up. ``` ``` I have a 30-day RDS GRI on no-upfront. What are my options when the lock-in ends? ``` ## 8. Moneyback Guarantee, Rebates, and Buybacks Tools: `get_moneyback_guarantee_savings`, `commitments_metrics` ``` How much has the Moneyback Guarantee saved me? Break it into rebates versus avoided commitment cost from releases and buybacks. ``` ``` Show my rebate history by month for this year. ``` ``` Explain how the moneyback guarantee would work if one of my 1-year GRIs went underutilized. ``` ## 9. Invoices, Finance, and Showback For anything that has to tie to the bill — reconciliation, a fiscal close, a number going to finance — ask for invoices. Invoices are the source of truth for commitment savings, Archera premiums, and rebates as actually charged. Metrics tools return estimates. Tools: `list_invoices`, `get_invoice`, `invoice_line_items`, `list_allocation_reports`, `get_allocation_report`, `get_allocation_report_periods` ``` What did I pay Archera in March? Use the invoice, not estimates. ``` ``` Pull my invoices for the last fiscal year and total net commitment savings by month. List which months you included. ``` ``` Break down last month's invoice line items by commitment, and show insured versus uninsured using the term breakdown. ``` ``` I need a showback report for last month: premiums, rebates, savings, and commitment cost allocated per account. Give it to me as a table. ``` ``` Which allocation reports exist, what periods are available, and what parameters does each accept? ``` > **Tip** — Say the word "audit," "reconcile," or "finance" in your prompt and Archera AI will reach for invoices first, then use metrics only for trend context. ## 10. Cost Attribution and Segments A segment is a saved filter over your cloud resources — the unit for team-level or environment-level attribution. Tools: `list_segments`, `get_segment`, `segment_filters`, `create_segment`, `daily_savings` ``` How are my current segments defined? ``` ``` What filters can I use to create a segment? ``` ``` Help me create a segment for everything tagged env=production in my two main accounts. Show me the definition before you create it. ``` ``` Show savings and coverage for the platform-team segment over the last 90 days. ``` ``` Which segment has the most uncovered reservable spend? ``` ## 11. Inventory and Price Catalog Two distinct datasets, and naming the right one saves a round trip: * What you are running (your resources, with trailing-30-day spend and usage) → `search_resource_skus`, `resource_sku_attributes`, `resource_sku_details`. * What is purchasable (the cloud's catalog, plus your custom-priced SKUs) → `search_catalog_skus`, `catalog_sku_attributes`, `catalog_sku_details`. ``` Which of my resources cost the most over the last 30 days? ``` ``` Which resources have the lowest commitment coverage? ``` ``` What services am I running, and which of them are reservable? ``` ``` What does this resource cost, is it reservable, and how long has it been running? ``` ``` Search the catalog for Azure Dsv6 SKUs in East US and show on-demand pricing. ``` ## 12. Product and Process Questions Archera AI searches Archera's documentation for product behavior — what the platform can and cannot do, how commitment types work, permissions, integrations, and the partner program. This is separate from your account data. Tools: `search_docs` ``` Does Archera support GCP CUDs, and which GCP services are covered? ``` ``` How does a 30-day Guaranteed RI actually work under the hood? What is Archera buying, and what am I locked into? ``` ``` What permissions does the AWS IAM role need for full commitment management, and what breaks if I only grant read-only? ``` ``` How do I connect an Azure tenant? Walk me through it step by step. ``` ``` Explain Archera's fee structure. When exactly am I charged a premium? ``` ## 13. Get Things Done in the App (In-App Only) These tools exist only in the Archera AI panel inside the Archera app. They are not available over Slack or through an external MCP client. Tools: `navigate_to_page`, `list_available_pages`, `open_new_commitment_plan`, `invite_user`, `contact_support` ``` Take me to the commitment inventory page for the commitment we just discussed. ``` ``` What pages can you navigate to? ``` ``` Open a new commitment plan for me. ``` ``` Invite my teammate to this organization. ``` ``` This needs a human. Connect me to support. ``` ## 14. Estimate Savings From a File You Have Attach a cloud cost export or invoice and ask for an estimate. Useful before your data has finished loading, or for an account that is not connected yet. Tools: `estimate_savings` (plus native reading of attached files) ``` [attach a cost export] Estimate what I could save with Archera based on this file. Show the assumptions you used. ``` ``` [attach a provider invoice PDF] Which line items here are reservable, and which are outside Archera's scope? ``` ``` [attach a CSV] Reconcile this spreadsheet against what my Archera invoice says for the same month, and list every discrepancy. ``` ## Limitations * **Archera AI cannot change your infrastructure.** Archera does rate optimization only — commitments are financial instruments. Right-sizing, instance-type changes, shutting down idle resources, and application changes are all out of scope. Archera AI will say so and suggest pairing Archera with a usage-optimization tool. * **It only sees what your role sees.** Access follows your Archera RBAC permissions. A permission failure is reported as one, not worked around. * **It will not invent numbers.** Archera AI quotes named fields from tool responses and does simple arithmetic on them (sums, differences, ratios, percentages) while showing its inputs. If a figure would require chaining several operations across fields, it says the exact value is not exposed and gives you the closest real fields instead. It also will not invent a provider-side mechanic to explain a coverage gap — if the data does not show the cause, it tells you that a deeper diagnosis needs the provider console. * **Estimates are labeled as estimates.** Metrics tools estimate from usage and can drift from the bill when discounts or credits are in play. Invoices are the source of truth. * **Cost data lags about 1.5 days, and end dates are exclusive.** Ranges are silently clamped to available data, and Archera AI surfaces the effective range when it materially changes the answer. * **Write actions need your explicit "yes."** Applying a plan, updating or deleting a plan, and any template write all pause for confirmation with a summary of what will change. * **Savings percentages in general explanations are population-level ranges**, not a rate for your specific SKU or account. Ask Archera AI to fetch your actual numbers. * **Context does not cross surfaces.** A Slack thread and an in-app conversation are separate. In Slack, the thread itself is the memory — start a new thread for a new topic, and note that a channel `@mention` sends the whole thread to the agent. * **No general web browsing.** Archera AI reads Archera's own documentation and your Archera data. It does not browse the internet. * **Ambiguous prompts get ambiguous answers.** "Compare to 3-year" without a payment option, or "last month" spanning a fiscal boundary, are the two most common sources of a technically-correct-but-unhelpful reply. Name the scope. ## Templates to Customize Replace the bracketed parts. **Investigate a spend increase** ``` My [AWS] spend went up about [$40k] between [June] and [July]. Break the change down by service, separate reservable from non-reservable, and tell me how much of it commitment coverage could have prevented. ``` **Decide on a commitment** ``` I have [$120k/month] of uncovered [EC2] spend. My workloads are [stable but we may migrate one service in Q4]. Recommend a plan, show the net monthly savings and breakeven, and be explicit about what I am locking into. ``` **Coverage audit** ``` Audit my commitment coverage for [AWS] over the last [90 days]. For each service: reservable spend, covered percentage, and uncovered monthly cost. Table, sorted by uncovered cost descending. ``` **Term-length what-if** ``` Take plan [plan name] and compare it against [3-year] [no-upfront]. Give me plan-wide totals, delta versus current, and how many line items would fall back to a shorter term. ``` **Finance close** ``` For [fiscal Q2], pull invoices for each month and report gross savings, Archera premiums, rebates, and net savings per month with a total. Say which months you included and flag anything that does not reconcile. ``` **Renewal review** ``` Everything expiring in the next [60 days]: what it covers, current utilization, and whether you would renew, resize, or let it lapse. One line of reasoning each. ``` **Showback** ``` Allocate last month's [premiums, rebates, savings, commitment cost] across [accounts] for [month/year]. Table, one row per account, sorted by net savings. ``` **Executive summary** ``` Write a five-bullet summary of my [AWS] commitment position for a non-technical executive: what we are saving, how well covered we are, the biggest remaining opportunity, and the main risk. No jargon. ``` --- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question. Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter: ``` GET https://docs.archera.ai/help-center/user-guide/archera-ai-prompt-guide.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.