ConnectMachine is a digital business card and contact management app from Connect Machine Inc. that scans paper cards, records meeting notes, and lets you search your network in plain language across iOS, Android, and any MCP-compatible AI client.
Used this tool? Rate it
Used this tool? Rate it
ConnectMachine is a digital business card and contact management application built around a simple premise: the hard part of networking is not exchanging details, it is remembering why the exchange mattered three months later. The product is published by Connect Machine Inc., a company that lists its address as 2261 Market Street STE 85698, San Francisco, CA 94114, and its marketing site leads with the line "Intros That Go Beyond Hello."
The company describes the underlying goal on its own terms as "The easiest way to exchange contact info with anyone and remember why you met them." That framing is worth taking seriously, because it explains the shape of the feature set. A conventional digital card app optimizes for the moment of the handshake. ConnectMachine adds a memory layer on top: voice notes attached to individual contacts, recorded meetings that become transcripts and summaries, automatic grouping of everyone met at the same event, and a natural-language agent that can answer questions about the resulting archive.
The positioning is explicitly not social. The company's about page describes the product as "A private, agentic network for meaningful connections, built by Connect Machine Inc." There is no public feed, no follower count, and no discovery graph. Your contacts are yours; other users cannot browse them. For a certain kind of professional — one who attends conferences, collects a stack of cards, and then never processes them — this is a meaningfully different value proposition from a social network that happens to store phone numbers.
Connect Machine Inc. is a US company. Its terms of service specify that "This Agreement will be governed by the laws of the State of California and the United States without regard to conflicts of laws provisions thereof," with disputes resolved in San Mateo County, California. The terms carry a "Last Updated on October, 15, 2025" date. The mobile app is published on both major stores under the seller name Connect Machine, Inc., and the marketing site, documentation, legal pages, blog, and comparison pages all live on the connectmachine.ai domain.
One small inconsistency is worth flagging up front because it recurs across the site: the homepage footer, about page, and privacy policy all use the spelling "Connect Machine Inc." (three words), while the terms of service uses "ConnectMachine Inc." (two words). These are almost certainly the same entity with a typographical drift rather than two companies, but anyone conducting formal vendor diligence should ask which spelling matches the certificate of incorporation.
ConnectMachine markets several AI capabilities — a scanner that reads business cards, a note taker that transcribes meetings, and a concierge that answers questions about your network. It does not, anywhere in its public material, disclose which underlying models or providers deliver those capabilities. A review of the homepage source, the terms of service, the privacy policy, the documentation, and the feature pages turns up no mention of OpenAI, Anthropic, Google, or any transcription vendor.
The closest the legal text comes is a subcontractor clause stating that "ConnectMachine may use the services of subcontractors for performance of services under this Agreement." That is a permission, not a disclosure. It is important not to misread the company's MCP integration as a model disclosure either: that feature lets ConnectMachine be controlled from AI clients such as Claude and Cursor, which is the opposite direction of dependency from ConnectMachine calling those vendors' models. Buyers with model-provenance requirements — regulated industries, or anyone whose procurement checklist asks where transcript audio is processed — should ask the vendor directly, because the public site does not answer it.
The central object in ConnectMachine is not a profile but a card, and the app is built around having several. The company's pitch is "Build a card for every context and choose exactly what to share. Nothing more, nothing less." In practice this means a founder card that leads with the company and a calendar link, a personal card that shares a mobile number, and a speaker card that leads with a bio — each drawing on the same underlying identity but exposing a different subset.
Cards can be branded with colors, a logo, and fonts. On the free tier the ceiling is four cards; paid tiers remove the cap. Cards are editable after the fact, and the company states that updates propagate: "Edit your card once, and ConnectMachine syncs changes everywhere automatically." That behavior is the practical argument for a digital card over a paper one — a job change does not invalidate every card already in circulation.
The most common failure mode for digital business cards is the friction they impose on the recipient. ConnectMachine's answer is a QR code that resolves to a web page: "They just point their camera. No download, no signup, your info lands straight in their phone." The recipient does not need an account.
Beyond QR, the app supports Apple Wallet and Google Wallet passes, AirDrop, SMS, email, and a generated email signature. The Wallet integration is the more interesting one in daily use, because a Wallet pass sits on the lock screen and can be surfaced without unlocking the phone and hunting for an app. There is also a "ShareBack" flow, in which the other party fills in their own details, which inverts the data-entry burden and tends to produce cleaner records than OCR.
The scanner handles the reverse direction — turning other people's paper into your structured data. The company describes it as "Our AI reads any business card and turns it into a clean, searchable contact, instantly." It handles paper cards, QR codes, and event badges, and the app description notes it also captures LinkedIn QR codes, saving the contact automatically.
The free tier permits up to three physical card scans in total, which functions as a trial rather than an ongoing allowance. Unlimited scanning is the headline unlock of the Individual plan. There is also a separate bulk scanning tool on the website for processing a stack at once.
The note taker records a conversation and processes it afterward. The documentation states: "Record your meetings and let ConnectMachine do the writing. When you stop recording, AI Notetaker processes the audio and gives you a clean summary and full transcript, ready to share." Each recording yields three artifacts: a full transcript, a short summary of key points, and speaker labels.
The feature is integrated with the contact database rather than standing alone. Per the documentation, "When you select a speaker who is already in your network, AI Notetaker links them to that contact, keeping your meeting notes connected to the right people." That linkage is what turns a pile of transcripts into a searchable relationship history.
The company advises transparency with the other party, framing it as good practice: "Let everyone know you are taking notes to keep things transparent." That advice is sensible but it is not a compliance control — recording consent law varies by jurisdiction and in many places is a legal requirement rather than an etiquette suggestion. Notably, the public material does not state a maximum recording length, the list of supported transcription languages, or where audio is stored and for how long.
The concierge is the natural-language layer over everything above. The company's description is "Talk to your contacts and calendar like a real assistant. It schedules, searches, and remembers for you." The documentation elaborates that it is "Your private AI agent that understands your network. Ask questions in plain language, recall who you met and where, and trigger actions like scheduling or sharing without digging through your contact list manually."
The example query style the company uses is instructive: "Who did I meet at the Austin conference?" This is the kind of question a contact list sorted alphabetically cannot answer, and it is only answerable because the app has been capturing event context, notes, and timestamps alongside the raw contact fields.
Contacts can be grouped into named networks, tagged, and filtered by event, company, or priority. A smart merge feature handles the duplicates that inevitably accumulate when the same person is scanned twice and shared once. The company states that "Both free and Premium include unlimited contacts, with organization tools like tags, networks, and search that scale into the thousands," which means the free tier's limits fall on cards and scans rather than on how many people you can store.
Contact enrichment fills gaps automatically; the documentation says "ConnectMachine automatically enriches the contact with publicly available data." Users evaluating this should note that automatic enrichment from public sources has data-protection implications in some jurisdictions, and the company does not name its enrichment sources.
The MCP server is the feature that most clearly separates ConnectMachine from mainstream digital card apps, and it is unusually thorough for a product in this category. The integration page describes it as the ability to "Manage contacts, search your network, and update your digital cards directly from Claude, Cursor, VS Code, Zed, and any MCP-compatible AI tool."
The server lives at the endpoint https://mcp.connectmachine.ai/mcp and authenticates with a bearer token. It exposes roughly forty tools spread across six groups: twelve for contacts (list, get, find, create, update, delete, smart search, industries, counts, duplicates, merge, network assignment), three for events, five for networks, seven for digital cards, six for import and export, and seven for notifications and profile. Plan gating carries over — as the documentation puts it, "Actions that are Premium in the app stay Premium over MCP, such as CSV export and filtering by industry or location."
One security detail deserves emphasis rather than burial: the documentation states that MCP keys "do not expire." A non-expiring bearer token with full read and write access to an entire professional network is a credential that requires deliberate handling. If it leaks through a shared config file, a synced dotfile repository, or a screen share, there is no built-in expiry to limit the blast radius, and the public documentation does not describe a rotation or revocation workflow.
CRM connectivity runs through Zapier rather than native connectors. The documentation names Salesforce, HubSpot, Zoho, and Pipedrive as supported destinations, with Notion, Airtable, and the rest of the Zapier catalog reachable through the same path. Webhooks are also available for custom integrations, with documented delivery semantics: a ten-second timeout, up to three retries spaced six minutes apart for four attempts total, and response bodies in logs capped at 4096 bytes.
The absence of native CRM connectors is a real gap rather than a pedantic one, and it has been noticed by users. On Product Hunt, the most substantive improvement request was for native integrations with major CRM platforms such as HubSpot and Salesforce. Zapier works, but it adds a subscription, a failure surface, and a latency step that a native connector would not.
This is the scenario the product is most obviously designed for. Over three days a delegate collects forty cards, and by the following week the faces have detached from the names. ConnectMachine's answer is to capture context at the moment of collection: scan the card, dictate a ten-second voice note, and let smart event detection group everyone met at that venue. The follow-up email a week later can then reference the actual conversation rather than opening with a generic pleasantry.
Consultants, realtors, insurance agents, recruiters, and sales representatives all share a pattern: many short meetings, a need for context on each, and a CRM that only gets updated when someone remembers to update it. The company maintains dedicated landing pages for each of these professions. The value here is the capture-to-CRM path — a scan flows into the contact database and, via Zapier, onward into Salesforce or HubSpot without a manual re-entry step.
Recruiters meet candidates at career fairs and campus events in volume, and the differentiating information is rarely on the card — it is the impression formed in the four-minute conversation. Voice notes attached to each contact, plus the ability to later ask which candidates were met at a specific event, addresses the part of the workflow that an applicant tracking system typically handles poorly at the top of the funnel.
Some professionals do not want a public networking profile at all. Doctors, dentists, and allied health practitioners — all of whom the company targets with dedicated pages — often need to share a professional contact without exposing a personal mobile number or a social profile. Selective sharing, where each card exposes only a chosen subset of fields, is the direct answer to that requirement.
The MCP server enables a workflow that is genuinely uncommon in this category: managing a professional network from inside an AI coding client. A user working in Claude Desktop or Cursor can ask for a contact's details, create a card, or run a smart search without leaving the tool. For people who already live in these clients, removing an app switch is a real efficiency gain, though this remains a niche within a niche.
The Team plan targets the case where a sales organization or agency wants every member's card to look the same. It offers centralized billing, seat assignment, team analytics, custom domain support, admin branding controls, and priority support. The problem it solves is brand drift — the tendency for individual employees to improvise their own formatting when left to their own devices.
Download the app from the App Store or Google Play, or start on the web. The company's own guidance is direct: "Download the app, sign up, and follow the guided steps to build your first digital card." Start with the identity you present most often. Decide deliberately what goes on it — the selective sharing model only pays off if you actually exercise the choice rather than filling in every field by reflex.
Once the primary card exists, add cards for your other contexts. The free tier allows four, which is enough to cover most people's real segmentation: work, personal, and one or two situational variants. Apply branding at this stage so that later edits are cosmetic rather than structural.
Add the card to Apple Wallet or Google Wallet. The company's instruction is to "tap 'Add to Wallet' from your card screen to access and share it instantly." This step matters more than it sounds: the difference between a card you can surface from the lock screen and one that requires opening an app determines whether you actually use it when someone asks. Home screen widgets and Control Center quick actions serve the same purpose.
When you receive a paper card, scan it immediately rather than pocketing it. Ration the free tier's three scans on cards that matter, or use ShareBack — having the other person enter their own details produces more accurate records than OCR and does not consume a scan. Point-of-capture discipline is the whole game; a card scanned three weeks later has lost the context that makes it valuable.
Immediately after the conversation, record a voice note or start the AI Notetaker. The company's framing is "Record conversations on the spot and attach them to the contact, so you always remember what you talked about." A fifteen-second note capturing what the person is working on and what you agreed to do next is worth more than any amount of enrichment data. If you record a full meeting, tell the other participants first.
Group contacts into networks and apply tags as they accumulate; let smart merge clear the duplicates. Then use the concierge rather than scrolling — ask who you met at a given event, or which contacts came from a specific company. This is the payoff step, and it only works if steps 4 and 5 were done consistently.
If you use a CRM, set up the Zapier connection so new scans flow into Salesforce, HubSpot, Zoho, or Pipedrive automatically. If you work inside an AI client, generate an MCP key and add the config. If you need a portable backup, export to CSV from settings — the company documents the path as "under settings > Export Contacts."
Treat the free tier as an evaluation, not a plan. Three lifetime scans is a trial allowance. If the scanner is the reason you are here, plan on the paid tier from the outset rather than being surprised on the fourth card.
Prefer ShareBack over scanning at high-volume events. It does not consume your scan allowance, and self-entered data is generally cleaner than OCR output. Reserve scanning for cards handed to you when the other person has already walked away.
Write voice notes for the future reader, not for yourself today. "Interested in the Q3 pilot, wants a follow-up after their board meeting in September" is retrievable. "Nice guy, good chat" is not. The concierge can only surface what you gave it.
Verify transcripts before forwarding them. Automatic speech recognition degrades with accents, crosstalk, and background noise — exactly the conditions of a conference floor. A summary sent to a client with a misattributed commitment is worse than no summary.
Get consent before recording, in writing where it matters. The company's "let everyone know" guidance is a floor, not a ceiling. Several jurisdictions require all-party consent, and the vendor's etiquette advice does not transfer legal risk away from you.
Treat your MCP key like a password, because it functions as one. It does not expire and it grants full access to your network. Keep it out of committed config files, out of synced dotfile repositories, and off screen shares.
Export a CSV on a schedule. Your professional network is data you accumulated over years, and it should not live in exactly one vendor's database. The export is available on the free tier, so there is no cost argument against doing it quarterly.
Decide your card taxonomy before you build cards, not after. Retrofitting a segmentation scheme onto cards already in circulation means the versions other people saved are stale.
High-frequency networkers get the most out of it. If you attend conferences, run business development, or meet people in volume, the memory layer solves a problem you genuinely have. The company positions the product for "Founders, investors, executives, speakers, agency owners, and sales teams."
Client-facing independents — consultants, freelancers, coaches, realtors, insurance agents — benefit from selective sharing and the capture-to-CRM path without needing enterprise tooling. Independent coverage in Dynamic Business made the same assessment, describing the fit as professionals who "need a reliable tool to capture not just contact info but context."
Small teams wanting brand consistency are served by the Team plan's admin controls and shared templates, at a price point well below enterprise sales enablement platforms.
Privacy-conscious professionals who want the utility of a digital card without a public profile are a deliberate target; the no-public-feed design is a product decision rather than a missing feature.
AI-native workers who already operate inside Claude, Cursor, or VS Code will find the MCP server genuinely useful, and it is unusual enough to be a deciding factor for that group.
Who it does not suit: anyone wanting social discovery or a public presence, since there is deliberately none. Anyone needing native CRM connectors today rather than Zapier. Anyone with strict procurement requirements around AI model provenance or data residency, given the disclosure gaps discussed below. And anyone who will not maintain the capture habit — an empty memory layer is just a slower contact list. The Dynamic Business assessment noted the same boundary, observing that users who prefer "public social media elements" may find it restrictive.
ConnectMachine ships as native mobile apps plus a web presence. The iOS app is listed as "CM: Digital Business Card," published by Connect Machine, Inc., at version 2.2.0 with a release date of 30 October 2025 and a most recent update on 1 August 2026. It requires iOS 15.5 or later, occupies roughly 147 MB, is rated 4+, and is categorized under Business and Productivity. Store metadata lists six interface languages: English, French, Italian, Russian, Chinese, and Spanish.
The Android app is published under the package identifier com.connect.machine by the same seller, showing a 4.7 rating, more than 10,000 downloads, and a most recent update dated 31 July 2026. Its in-app purchase range is listed as $4.99 to $199.99 per item, a wider band than the headline subscription pricing alone would suggest.
Wallet integration covers both Apple Wallet and Google Wallet, including lock-screen presentation on iOS. Recipients need no platform at all — a QR scan opens a web page in any phone's browser. The marketing site itself is published in five languages (English, Spanish, Russian, Italian, and French), which is a narrower set than the six the mobile app claims, so language coverage depends on which surface you are using.
For automation, the Zapier integration and the MCP server extend reach to Claude Desktop, Claude Code, Cursor, VS Code, Zed, Windsurf, and Cody, along with any other MCP-compatible client that can reach a remote HTTP server.
ConnectMachine publishes its pricing openly on the homepage, which is worth noting because many tools in this category require a sales conversation to learn the number.
Free — $0, described as "Forever." The company summarizes it as "Basic access with essential features." It covers up to four cards, up to three physical card scans, CSV contact export, Apple and Google Wallet integration, a WhatsApp assistant, and home screen widgets.
Individual — $5.99 per month. The main unlocks are unlimited physical card scans, unlimited cards, and CRM integration. The company's comparison material states the annual option as "$5.99 per month, or $59.99 per year," and the homepage advertises "Save 20% on yearly."
Team — $5.99 per seat per month. Positioned as "Same Premium features. One dashboard," it adds centralized billing, seat management, team analytics, custom domain support, admin branding controls, and priority support. Team signup routes through a "Book a Demo" flow rather than self-service checkout.
There is no credit system or metered consumption model. Access is a flat subscription, with the free tier's constraints expressed as hard counts on cards and scans.
Two caveats belong with the pricing. First, the homepage's own plan table is internally inconsistent: the Free column lists AI Notetaker, AI Concierge, and Netsend among its inclusions, while the Individual column simultaneously presents those same three items under "Everything in Free, plus." Both cannot be true. The company's comparison page resolves it in the direction of the paid tier, stating the AI Notetaker is "Included on Individual." Prospective users relying on the AI features should verify their availability on the free tier before assuming access.
Second, the refund position is strict. The terms of service state plainly: "All fees and payments made by Customer to ConnectMachine are non-refundable." Subscriptions renew automatically, and the terms require "thirty (30) days' prior written notice" for cancellation — a longer window than the 24-hour-before-renewal convention that governs App Store purchases, which is itself the mechanism actually applied to mobile subscriptions. Where the two conflict, the store's rules generally govern store-billed subscriptions, but the discrepancy is unresolved in the vendor's own documents.
Blinq is the most directly compared competitor, and ConnectMachine's own comparison page lays out the numbers: Blinq at "$9.99 per month, or $7.33 per month billed yearly (about $88 per year)" against ConnectMachine's $5.99. Blinq's team pricing is cited as "$6.99 per card per month ($4.99 billed yearly), 5-card minimum." Blinq's advantages, per that same page, include native CRM sync from the free tier and a broader integration catalog on its business plan — precisely where ConnectMachine is weakest. Blinq also sells NFC hardware, which ConnectMachine does not.
HiHello and Popl are the other two competitors the company maintains comparison pages for. Popl is strongly oriented toward NFC hardware and team lead capture at events; HiHello is a mature digital card product with well-developed card design and management. Both are reasonable choices for buyers who want a card product without the memory layer.
Native CRM mobile apps from Salesforce or HubSpot cover the capture-to-pipeline path directly and without an intermediary, and for a sales team already standardized on one, that may be sufficient. They are, however, much heavier tools, and they do not produce a shareable personal card.
Dedicated meeting recorders such as the various AI notetaker products handle transcription with more depth and typically publish their language support and retention policies. If transcription is the primary need and the card is incidental, a specialist tool is the better fit.
Paper cards remain a legitimate option in contexts where they are the expected convention. The company's own blog engages this question directly rather than dismissing it.
The honest summary: ConnectMachine competes on price, on the memory layer, and on the MCP integration. It does not compete on native CRM connectivity, integration breadth, or the transparency of its AI supply chain.
No disclosure of AI models or providers. This is the most significant gap. Across the homepage, terms, privacy policy, documentation, and feature pages, the company never states which models or vendors power the scanner, the transcription, or the concierge. The terms reserve only a generic right that "ConnectMachine may use the services of subcontractors for performance of services under this Agreement." For most individual users this is immaterial; for regulated buyers it is a blocking question that must be asked directly.
Conflicting statements on data residency. The App Store description asserts "GDPR compliant, EU data stays in EU datacenters," while the privacy policy states "Our Services are hosted in the United States." These claims are difficult to reconcile as written. The privacy policy is the controlling document, and EU-based buyers should treat the store copy as marketing until the vendor confirms otherwise in writing.
Softer language on data sale in the binding document. The homepage FAQ says the company will "never sell your data." The privacy policy says "We do not sell your Personal Information in the traditional sense." The qualifier "in the traditional sense" is a meaningful hedge under CCPA, where "sale" has a statutory definition broader than an exchange for cash.
AI training rights are asserted, with a de-identification condition. The terms state that "ConnectMachine may use Inputs and Outputs to train or otherwise improve the AI Services, but only if such Inputs and Outputs have been (a) de-identified." Meeting recordings and contact notes are Inputs. The privacy policy contains no corresponding training disclosure, so the two documents describe the arrangement at different levels of detail, and there is no documented opt-out.
Non-expiring MCP credentials. Convenient, but a durable full-access token with no documented rotation path is a real exposure for anyone whose network is sensitive.
Undocumented limits on the recording feature. No maximum recording duration, no list of supported transcription languages, and no stated retention period or storage location for audio appear in the public material. Transcription accuracy on accented or noisy speech is likewise unstated and should be tested before relying on it.
CRM integration is indirect. Everything routes through Zapier, adding cost and a failure point. This was the leading improvement request from users on Product Hunt.
Traction claims are inconsistent across surfaces. The homepage cites "Loved by 22K users with a ⭐ 4.7 rating" and elsewhere "4.8 avg. rating" with "96% 5-star." Verifiable store data is more modest: the iOS listing carries a 5.0 average from six ratings, Google Play shows 4.7 with over 10,000 downloads, and the Product Hunt listing shows 4.3 out of 5 from three reviews. The Android download figure is the only independently checkable scale indicator, and it does not by itself substantiate 22,000 users.
Thin independent review coverage. There is no G2, Capterra, or Trustpilot presence, and no substantive Reddit discussion — searches for the latter return same-named but unrelated products. Independent coverage amounts to a Dynamic Business writeup and a small number of Product Hunt reviews, which is a thin evidentiary base for a purchasing decision.
Entity name inconsistency. "Connect Machine Inc." on most pages versus "ConnectMachine Inc." in the terms of service.
Uneven localization. Six app languages against five website languages, and the Dynamic Business assessment noted that "some integrations or language support are nascent."
Yes, there is a genuinely free tier priced at $0 and described by the company as "Forever." It includes up to four cards, up to three physical card scans, CSV export, Apple and Google Wallet integration, a WhatsApp assistant, and home screen widgets. The three-scan allowance is a lifetime total rather than a monthly refresh, so the free tier works best as an evaluation. Note that the homepage plan table is contradictory about whether the AI Notetaker and AI Concierge are included at no cost — the comparison page indicates they belong to the Individual plan. Verify current availability in the app before depending on them.
The Individual plan is $5.99 per month, with an annual option the company states as $59.99 per year and advertises as saving 20%. The Team plan is $5.99 per seat per month and is sold through a demo request rather than self-service. Google Play lists a wider in-app purchase range of $4.99 to $199.99 per item, which suggests additional purchase options beyond the two headline tiers. Be aware that the terms state all fees are non-refundable.
The company does not say. No model or vendor is named anywhere in its public material — not on the homepage, not in the terms, not in the privacy policy, and not in the documentation. The terms mention only that subcontractors may be used. If your organization requires knowing where meeting audio is processed or which provider handles inference, you will need to ask the vendor directly. Do not interpret the MCP integration as an answer: that feature lets AI clients like Claude control ConnectMachine, not the other way around.
No. Sharing is designed so the recipient needs nothing installed. As the company puts it, "They just point their camera. No download, no signup, your info lands straight in their phone." A QR scan opens a standard web page. Apple Wallet, Google Wallet, AirDrop, SMS, and email sharing are also available, and you can generate an email signature carrying the same card.
Indirectly. There are no native CRM connectors; integration runs through Zapier, with Salesforce, HubSpot, Zoho, and Pipedrive named in the documentation, plus the wider Zapier catalog including Notion and Airtable. Webhooks are available for custom builds, with a documented ten-second timeout and up to three retries six minutes apart. CRM integration is a paid-tier feature. Native connectors were the leading improvement request from users on Product Hunt.
You start a recording, and when you stop, the app processes the audio into a full transcript, a short summary, and speaker labels. Speakers already in your network are linked to their contact records automatically. The company advises telling participants you are recording. Important gaps: the public material does not state a maximum recording length, which transcription languages are supported, or where recordings are stored and for how long. Recording consent is legally regulated in many jurisdictions, so confirm the rules that apply to you rather than relying on the vendor's etiquette guidance.
MCP lets you drive ConnectMachine from AI clients including Claude Desktop, Claude Code, Cursor, VS Code, Zed, Windsurf, and Cody. The server is at https://mcp.connectmachine.ai/mcp and exposes about forty tools covering contacts, events, networks, cards, import and export, and profile data. You authenticate with a bearer key, and premium gating carries over — CSV export and filtering by industry or location remain paid features over MCP. One caution: the documentation states these keys do not expire, so treat the key as a long-lived credential and store it accordingly.
Contacts are private to you by default, with no public feed and no follower mechanics, and each card exposes only the fields you choose. You can export contacts to CSV from settings, and this is available on the free tier. You can delete your account from app settings. Two caveats deserve attention: the privacy policy says data is not sold "in the traditional sense" rather than flatly not sold, and the terms permit using de-identified inputs and outputs to improve the AI services, with no documented opt-out.
The company states GDPR compliance and the privacy policy acknowledges GDPR rights for EEA and UK users and CCPA rights for California residents. Data residency, however, is contradictory in the source material: the App Store description claims EU data stays in EU datacenters, while the privacy policy states that services are hosted in the United States. The privacy policy is the binding document. EU organizations with data residency obligations should obtain written confirmation before proceeding.
Connect Machine Inc. is based in San Francisco, with terms governed by California law and last updated in October 2025. The iOS app was released in October 2025 and reached version 2.2.0 by August 2026, indicating active development. Independently verifiable adoption is modest: over 10,000 Android downloads with a 4.7 rating, a 5.0 iOS rating from six ratings, and a 4.3 Product Hunt rating from three reviews. The homepage's claim of 22,000 users is not corroborated by these public figures. There is no G2, Capterra, or Trustpilot presence, so independent review evidence is limited.
