"Information about Amazon Web Services (AWS)" matching MCP connectors:

Scores web domains: TLS, headers, DNS, speed. Free triage, $0.01 per report, no signup.

# Customer Intelligence & Segmentation **Segment your book. Predict what each customer does next. Price it in dollars.** Most tools sold as "customer intelligence" are descriptive segmentation with a better chart — they tell you what your customers already did. This one segments your book, forecasts what each customer will do next, and attaches a dollar figure with a confidence range to every forecast. Input: three columns — `customer_id, date, amount`. Available from Excel, Salesforce, HubSpot, Zoho or any accounting export today, no integration project required. Full results in under two minutes at any book size, on SOC-compliant infrastructure — and free while we're in early access. --- ## The two halves, and why both matter **Segmentation is the map. Intelligence is the decision.** **The segmentation layer** answers: who are my customers, grouped so I can act? It outputs A/B/C/D tiers, RFM behavioural cells, revenue concentration (Gini), and named buckets like Champions and At-Risk. It looks backward at what already happened. **The intelligence layer** answers: what will each one do next, and what's it worth? It outputs survival probability, predicted lifetime value, migration risk, value-at-risk, contact uplift, and total book exposure. It looks forward. Both refresh nightly. Segmentation alone produces a label. A label is a classification, not a decision — which is why CRM tier fields sit unused in every company that has them. The intelligence layer converts the label into an action and a number, and it's where this product actually lives. A concrete example of the difference. Segmentation says: *"Acme is an A-account."* Intelligence says: *"Acme is worth $60,000/yr, its purchase cadence has broken, lapse probability this quarter is 35%, that's ~$21,000 at risk ($16,000–$27,000, 90% CI), and a contact this week recovers an expected $14,000 of it."* Only one of those changes what someone does on Monday. --- # For the C-suite — CEO, CFO, Founder ### What's actually broken **The board question has no defensible answer.** "How much revenue is at risk this quarter?" is currently assembled by hand, a month stale, and nobody can reconstruct how it was calculated. **Concentration risk is invisible until it detonates.** If four accounts carry 60% of revenue and two have gone quiet, that fact is hiding inside a perfectly normal-looking dashboard. **You bought reporting and hoped for intelligence.** Dashboards describe the past. Enterprise-grade predictive scoring sits behind expensive editions plus a data-science setup — structurally priced away from a mid-market business with a clean CSV. **The alternative is headcount.** A data analyst runs $90,000–$130,000 a year and still cannot recompute the entire book every night. ### What you get **One number for the board, with its working attached.** Something like: *"$340,000 at risk across 23 neglected accounts. $520,000 of upgrade upside sitting in 41 B-accounts."* Produced by `book_exposure(book_id)`, and every dollar inside it traces to a named customer, a named model, and a training date. **Concentration health as a tracked metric.** A Gini coefficient on your revenue book alongside the A/B/C/D split — "we're too dependent on our top accounts" stops being a worry and becomes a number with a trend line. **Segments built on where customers are going, not where they've been.** Tiers are assigned on *predicted* lifetime value, not last year's invoices. Telling you who **will** be cream rather than who **was** is the entire product claim. **Numbers built to survive being wrong.** Every dollar figure ships as a range with the model's real confidence — *"~$42,000 at risk ($31,000–$54,000, 90% CI)"* — never a bare point estimate. This is deliberate. A manager shown a confident "$42,000", who watches the account churn anyway and then learns it was a rough guess, stops trusting every number you've ever shown them. A range plus a reason survives the miss. **Security and compliance your procurement team can sign off.** SOC-compliant infrastructure and controls across the MCP portfolio, so the security review isn't the thing that stalls a rollout. Transaction data is personal data under most privacy regimes — consent capture, retention policy and delete-on-request are built in for GDPR and DPDP-style requirements. **Answers in under two minutes, at any book size.** Not a batch job you queue and check tomorrow. In testing, a full book scores in under two minutes regardless of how many customers are in it — so "what's at risk right now" is a question you ask during the meeting, not after it. ### What it costs today **Nothing.** The full product — every tool, every segment, the complete intelligence layer, unlimited book size — is free while we're in early access. No card, no seat count, no feature gating. We'd rather have your book running nightly and your feedback on the numbers than a purchase order. Pricing arrives later; early users get told well in advance and won't be switched off mid-quarter. --- # For VPs — Sales, RevOps, Customer Success ### What's actually broken **Reps work the accounts they like**, not the accounts that move the number, and you have no defensible way to argue otherwise on a Monday morning. **Segments don't change behaviour.** "This is an A account" tells a rep nothing about what to do today. **The highest-leverage moment is invisible.** A customer migrating between segments — a B slipping toward C, a C climbing toward B — is where one call changes the outcome. You currently learn about it from the revenue chart, one quarter late. **You can't prove coverage works.** When finance asks whether rep outreach actually retains revenue, you have anecdotes. **QBR prep is spreadsheet archaeology**, rebuilt from scratch every quarter with slightly different logic. ### What you get **A daily queue ranked by money, not by segment.** This is the most important design decision in the product. Value of contact = customer lifetime value, multiplied by the difference between the probability they stay if contacted and the probability they stay if not. An A-account that stays loyal no matter what has near-zero uplift — it does not belong at the top of anyone's call list. A B-account a single call would visibly save has high uplift and belongs at the top. Segment-ranked queues systematically spend your team's best hours on accounts that were never going anywhere. `contact_uplift()` fixes that. **Segment migration, watched nightly.** `watch_migrations(book_id)` recomputes the whole book overnight and alerts on slips and jumps. That alert list is the morning queue, delivered into Teams or Slack — not another tab nobody opens. **Fast enough to use live.** A full rescore returns in under two minutes at any book size. A rep can pull a fresh read mid-pipeline-review, and you can re-run the book after a bad month instead of waiting for the next scheduled refresh. **Growth targeting, not just save-the-churn.** `rank_uplift_targets(tier)` answers "which of my B's are most likely to reach A if we touch them," using a migration model estimated from two transition matrices: accounts your reps touched, and accounts they didn't. That same comparison is your proof to finance that coverage works. **Explanations reps believe.** `explain_tier(customer_id)` returns the whole trace — the ABC baseline, the RFM cell, the survival probability, the migration risk, the model version, the training date. A rep who reads *"recency score collapsed from 5 to 2 in 60 days"* picks up the phone. A rep handed an opaque score from a black box ignores it, and the deployment dies quietly. **Two framings from one engine.** Urgency: *"Acme is worth $60,000/yr; 35% lapse risk this quarter, so $21,000 is at risk."* Growth: *"Invest an hour in Beta Ltd and unlock $18,000 of upside."* **Book and territory exposure on tap** for QBRs, headcount arguments and comp design — computed identically every quarter, so the trend is real rather than an artifact of whoever built this quarter's spreadsheet. ### The toolset **`score_customer_book(source)`** — the full segmented ledger across the book. **`watch_migrations(book_id)`** — who moved overnight, which becomes today's queue. **`contact_uplift(customer_id)`** — who to call first, ranked by dollars. **`rank_uplift_targets(tier)`** — which B's can be grown into A's. **`value_at_risk(customer_id)`** — what neglect costs on this account. **`book_exposure(book_id)`** — the number you report upward. **`explain_tier(customer_id)`** — why, auditable and versioned. **`predict_customer_value(customer_id, horizon)`** — expected future spend over a horizon. **`data_quality_report(source)`** — dedupe and gap detection before anything is scored. --- # For agent builders — developers, ISVs, RevOps consultants ### What's actually broken **An LLM cannot fit a survival model.** It can rank a CSV once, plausibly, and be wrong in a way nobody catches. Ask twice, get two answers. Ship that to a client and you own the outcome. **Rolling your own is a trap with non-obvious failure modes.** BG/NBD maximum-likelihood fitting breaks specifically: log-Gamma overflow, flat likelihood ridges, degenerate books that converge silently to nonsense. **No audit trail means no enterprise sale.** "The model said so" ends the conversation when a client asks why an account was segmented as C. **Most vendors ship a UI, not primitives.** You need composable tools to wire into an agent, not a dashboard to iframe. ### What you get **Nine deterministic MCP tools** over a standard interface. Same input, same model version, same answer — every time. Structured outputs designed for an agent to consume, not a human to squint at. **Hardened numerics, not a thin wrapper.** Built on the `lifetimes` OSS lineage and hardened for production: parameter bounds, multi-start optimization, convergence diagnostics, golden tests, and refuse-with-reason below data thresholds. When a book is too thin or too degenerate to model, you get an explicit refusal with a reason code — not a confident hallucination your client discovers in front of their board. **Everything versioned.** `model_version` and `trained_through_date` on every response, plus the full explanation trace. Reproducible and defensible six months later. **Cross-source by design.** CSV/XLSX first, so you can demo without an OAuth dance. Dataverse Web API, Salesforce REST/Bulk, HubSpot and Zoho read-only connectors as they mature. Native CRM scoring only ever scores its own CRM; real businesses keep data in an ERP and a CRM and Excel. **Confidence intervals as first-class output.** If you're building an agent that recommends actions, you need to know when the model is guessing — and so does whoever receives the recommendation. **Sub-two-minute response at any book size.** Model fitting is the expensive part of this problem class, and naive implementations degrade badly as the book grows. Ours returns a full scored book in under two minutes in testing, independent of customer count — so you can call it inside an agent loop instead of architecting around a long-running job, a polling endpoint and a callback. **SOC-compliant infrastructure and controls** across the MCP portfolio. If you're embedding this in something you sell, your customer's security questionnaire becomes a form you fill in rather than a project you run. ### The models underneath **Layer 1** — Pareto / ABC classification plus Gini concentration. Juran / GE, 1951. **Layer 2** — RFM behavioural scoring into named segments. Hughes, 1994. **Layer 3** — BG/NBD survival modelling: probability a customer is still active, and expected future purchases. Fader, Hardie and Lee, 2005. **Layer 4** — Gamma-Gamma spend model into discounted customer lifetime value. Fader and Hardie, 2013. **Layer 5** — Markov migration and uplift, estimated from touched versus untouched accounts. Pfeifer and Carraway, 2000. **Layer 6** — Assembly, explanation trace, and versioning. Layer 4 runs a published validity check most implementations skip: the correlation between purchase frequency and average order value must be near zero, and it is tested and reported, because Gamma-Gamma is invalid without it. **For consultants:** run multiple client books, white-label the reports, and repeat the engagement shape — score the book, present the exposure, run the queue, show the migration delta at the next review. Currently free, with no cap on the number of books. --- ## Where this sits in your stack Worth being precise, because the category name invites the wrong comparison. **It is not a CDP.** It doesn't unify identity, ingest event streams, or store profiles. It reads transactions and returns decisions. **It is not a CRM replacement.** Your CRM stays exactly where it is; this reads from it and writes decisions back into your team's workflow. **It is not a BI tool.** BI describes what happened and leaves interpretation to you. This forecasts what happens next and prices it. **It is not a campaign or marketing automation platform.** It tells you who to reach and what reaching them is worth. Sending is somebody else's job. The closest honest description: **a predictive segmentation and account-prioritisation engine that sits between your transaction data and your team's daily work.** --- ## What it won't do Honesty is part of the product, so, plainly: **It doesn't model pure-subscription books out of the box.** The survival models assume non-contractual purchasing where customers can lapse silently. Subscription and renewal books need the documented renewal-mode variant — the engine says so rather than quietly mis-modelling you. **It doesn't manufacture answers from thin data.** Below the data threshold you get an explicit refusal with a reason, not a fabricated number. **It doesn't give you a single confident figure with no range.** If that's what you want, this product will disappoint you on purpose. --- ## Start in five steps 1. Export `customer_id, date, amount` from wherever it lives. 2. Run `data_quality_report(source)` to dedupe and flag gaps before anything is scored. 3. Run `score_customer_book(source)` for the full segmented ledger in one call. 4. Run `book_exposure(book_id)` for the number that goes to the board. 5. Turn on `watch_migrations(book_id)` so the nightly recompute becomes tomorrow's queue. Steps 2 through 4 take under two minutes each, whatever the size of your book. Free while in early access, with no customer limit.

Agent intake desk for property renovation. Read only, CORS open, fail closed, receipts recompute.

SSRF-safe web intelligence: 4 outcomes, 43 paid x402 tools, 3 free tools and a URL preview.

Read-only Amazon SES observability: search events, inspect bounces, pull delivery stats.

Meta-search MCP for x402 agent-payable services -- find paid APIs across every registry.

CatchAll is a web search API built for comprehensive event retrieval — not ranked results, but all matching records.

Docs: https://docs.keenable.ai/mcp-server Keenable is a free, remote MCP server that gives agents access to the web index. Search the web with ranked results and date/site filters, then fetch any indexed page as clean markdown. Works out of the box with no account or API key.

Find the stories worth writing about: discover emerging stories, rank the angle, draft from sources.

Give your agent web search and authoritative datasets: S&P Global, FRED, OECD, SimilarWeb & more.

Ask about your Dreambooth Studio photobooths: sessions, revenue, credits, projects, device status.

Scrape, crawl and search the web for AI agents via MCP.

SaaS intelligence for AI agents. 5 unified tools cover 1,000+ services with 91-96% token savings.

Read-only AWS cost analysis: find idle and underutilized resources, with evidence.

Brand health & trend footprint across PSFK expert graphs with citable sources, not web summaries.

Deploy a multi-user web app from your agent: hosting, auth, database, and permissions.

Fast, intelligent web search and web crawling. New mcp tool: Exa-code is a context tool for coding

X-Z-CS AI validation with auditable reasoning. 8 active tools; 5 in maintenance. v0.5.58

Get recommended by Amazon's AI. Hosted MCP server for Amazon listing compliance & generation.

Unauthenticated OAuth2 callback collector for Tollbooth MCP services