Mandit GlobalHouston · Pune · Since 2018

AI-powered demand.
Human-verified revenue.

A global B2B lead generation and demand generation partner. Nine live applications, one pipeline — from first-party signal capture to a booked meeting with a verified decision maker.

0
Live products & tools
0
Technographic categories
0
Vendours detected
0
Intent tiers
Nine applications running live on production infrastructure
MG
Intent Engine
First-party behaviour
Licensed intent feeds
Identity resolution
Suppression & consent
Human-verified revenue
Product surface

Four services. Three public tools. Nine live applications.

Everything below is running production code today — not a roadmap. The services feed the tools; the tools feed the applications; the applications feed the pipeline.

Services

B2B Demand Generation
Awareness and sales-ready demand across the buying journey.
Intent-Based Marketing
Reach buyers while they are still actively researching solutions.
Waterfall Content Syndication
Premium content distributed across vetted publisher networks, stage by stage.
Email Nurture
Personalised sequences that keep qualified leads moving forward.

Public tools

TechStack Lookup
A company's technology stack, keywords and buying intent from one domain.
Buying Intent
Topic intent scores, source attribution and live signal windows.
Email Launchpad
Free pre-flight check: spam risk, rendering and send timing.
My Account
Usage, daily allowance, plan and payment history in one place.

Applications

Mandit Timeline
Multi-tenant prospect timeline, pixel events, identity resolution and public share views.
Campaign Orchestrator
Campaigns, AI agents, RAG memory, omnichannel, schedules and analytics.
Suppression Connector
Seven-stage enrichment: validation, suppression, firmographic and technographic scoring.
Email & Phone Validators
Standalone ZeroBounce-backed email and carrier/DNC phone validation services.

Operator tools

MailPilot CRM
Campaigns, roles, landing page versions and per-campaign actions.
Score Tracker
Per-user shifts, campaign scoring, admin panel and a full activity log.
Landing Page Cloner
Clones competitor landing pages, rebuilds them locally and publishes versioned variants.
MAILSIM
Predictive campaign lab: competitor and intent intelligence, CSV enrichment, simulation.
The complete funnel

Eight stages. One account. Every step auditable.

Select any stage to see what actually happens inside it — which code runs, which data it produces, and what it hands to the next stage.

01
Signal collection
First-party site behaviour + third-party research feeds
02
Identity resolution
Anonymous visitor one account ID, deduplicated roles
03
Audience selection
14 firmographic and contact-level filters, live reach estimate
04
Prospect profiling
Company + person lookup, technographics, employment context
05
Validation & suppression
ZeroBounce, DNS, RDAP, libphonenumber, DNC windows
06
Scoring & status
0–100 lead score, qualified / usable / review / restricted
07
Activation
Email, WhatsApp, SMS, LinkedIn, voice — consent-gated
08
Feedback loop
Opens, clicks, replies and meetings become the next signal
Stage 01

Signal collection

Loading…
What happens
    Powered by
    Prospect profiling · interactive

    Every record is scored. The scoring is visible.

    Toggle the checks a pipeline ran. The score, the notes and the resulting status update live — this is the same scoreLead() function the Suppression Connector runs on every upload.

    0/100
    No data
    review
    Scoring notes
    Intent engine

    Intent is found in four tiers — and never faked quietly.

    Select a tier to see its worked example, the proof it produces, and the sources it can be traced back to. The overall score is a weighted blend: search 20% · website 30% · content 25% · purchase 25%.

    Weighted intent score

    0
    Overall intent
    score
    —
    —
    Every signal carries its source
    Review-site comparison page
    Industry publication article
    Directory listing view
    Job post (hiring signal)
    Vendor documentation page
    Intent is the one thing we will not fake quietly. Technographics can be inferred from a site. Buying intent cannot — it comes from licensed co-op publisher networks and review-site panels, and it is metered on every lookup.
    Technographics

    Twelve categories. Every vendor carries evidence.

    Select a category to see the vendors detected inside it, how the detection is evidenced, and how confident the engine is. 12 categories × 12 named vendors = 144 tracked technologies.

    —

    Evidence behind the detection

    Confidence —
    Competitive threat by scale archetype
    Enterprise high
    Mid-market medium
    Startup watch

    Stack categories per archetype: EMAIL · CRM · AUTOMATION · ANALYTICS · CDP · CMS · SUPPORT · ADS. A peer set is estimated per company, then the overlap against our own stack is scored.

    Audience selection · interactive

    Fourteen filters. The reachable count updates as you choose.

    Firmographic filters cut companies and contacts together. Contact-level filters cut the contact pool far harder. Select filters to see the deterministic estimate — same selections always produce the same number.

    Reachable contacts
    528,000,000
    100% of base database
    59,000,000 companies
    surviving
    Filter breakdown
    Website monitoring · MG Track

    We watch what a prospect does on the site — not just that they arrived.

    A self-hosted WordPress plugin captures page views, mouse paths, click positions, PDF reading time and the network owner of the IP — then reconciles it against email events into one person. No Google Analytics, no third-party handoff.

    Anonymous visitor RDAP network owner IP-matched email event One prospect record
    Page views
    IP, approximate city / region / country, ISP, company or network owner from the regional internet registry (RDAP) with ASN and reverse-DNS hostname, plus OS, device type, browser and version, language, referrer and landing URL.
    One view per visitor per page per window — reloads never inflate the count. Bots detected and ignorable.
    Mouse path & heat dots
    The cursor is sampled on the page (250 ms default) and stored in small batched requests. Each visitor's movement is drawn as a path with heat dots — green start marker, red end marker — over a live copy of the page itself, so you see what was actually under the cursor.
    The page behind the path loads with a marker URL excluded from tracking, so opening a visitor never counts as a visit.
    Link clicks, pinned in place
    Every link click is pinned on the map as a numbered marker at the exact spot it happened, with the link text and time on hover.
    Click events land in the same events table as the mouse path.
    PDF reading time
    PDF links are rewritten to open through a built-in PDF.js viewer — the download attribute is removed and the file is rendered to a canvas, so the browser never hands it to a plug-in that would save it. Reading time and mouse movement over the document are recorded.
    If PDF.js cannot load, the viewer falls back to an embedded frame: the open and the time are still recorded, mouse movement inside the document is not.
    One person stays one row
    Every visitor gets an anonymous random first-party cookie ID and a 30-minute session. When a visitor arrives without their cookie, an IP that is already known is carried forward instead of starting a second entry — and a Combine visitors with the same IP action folds existing duplicates into the entry seen first.
    Registry lookups are cached for one week. Both behaviours are switchable in settings.
    Visitor replay
    pricing page · 3 visits in 7 days
    pricing Start free trial 1 2
    Start marker End marker Heat density Cursor path Numbered click

    Where it runs

    Two dedicated tables — page views and events (mouse paths, clicks, PDF sessions) — both pruned automatically by age and a hard row cap. Nothing is written to post meta. A companion dashboard, MG Track Insights, shows any number of websites in one place, connected by app name and app password, and the export only ever reads.
    AI agents

    Agents do the reading. The ledger stays true.

    Every figure an agent moves also moves the funnel behind it — the enrichment ledger, the outreach queue and the counters are the same number. Select a family to see how it runs.

    Local voice agent — fully offline
    ASR, LLM and TTS all running on-device. No cloud call, no audio leaves the machine.
    ASR
    Qwen3-ASR 0.6B · MLX
    0.079
    LLM
    qwen2.5:7b · Ollama
    0.16s ttft
    TTS
    Maya2 Native 8-bit · MLX
    0.51s ttfa
    3.60s
    Naive end-to-end
    1.72s
    Pipelined
    0.94s
    + intent cache
    Speak the first sentence, let the rest follow — first clause to audio.
    Activation & programmatic

    One account, six channels, one suppression truth.

    Consent, suppression and DNC are applied on every delivery cycle — not cleaned up afterwards. Paid and owned distribution runs on the same audience definitions as the outbound programme.

    Suppression & compliance

    Suppression windows of 30, 60 and 90 days are evaluated against every record, and the UI reports exactly which window a hit landed in.
    • DNC flagged costs a record 40 points and forces a restricted status — it can never reach a rep.
    • Disposable domains cost 25 points and are marked invalid outright.
    • Role-based and shared inboxes are flagged or removed before delivery.
    • Publisher consent statements captured at the point of download; opt-in timestamp, source page and campaign reference recorded.
    • STOP keyword suppression is applied across every channel, not per channel.
    • GDPR, CCPA and Canada's Anti-Spam Legislation respected.

    Waterfall syndication

    A waterfall is not one campaign blasted everywhere. It is a sequence of publishing partners added in order, where each new stage is only unlocked once the previous one has earned it.
    1Publisher impressions48,000
    2Content downloads3,200
    3ICP matches940
    4Qualified leads delivered610
    • Stage one runs on a small panel of tier-one publishers — the job is to answer whether the asset pulls the right people, job titles and company sizes.
    • Stage two adds mid-tier publishers covering adjacent topics; stage three broadens across regions and reader segments.
    • Every new partner is measured against the stage-one benchmark: match rate, seniority mix, bounce rate, cost per qualified lead.
    • Quality gates: firmographic match, role and seniority, verified business email, consent and compliance — nothing passes a failed gate.
    Where we are honest: we have not run a paid media DSP on our own properties — no ads.txt, no Meta Pixel, no Google Ads conversion IDs. Our growth model is organic + outbound + LinkedIn content. We build and operate the publisher-side distribution and the first-party measurement, and we do not claim an ad-buying layer we do not run.
    Social & publishing

    We publish one post, on the day it is relevant.

    A post written on Monday about Monday's news and sent on Friday is already stale. The content is written the morning it ships, from research done that morning.

    Where we publish

    The daily post pipeline

    1
    Current research
    Same-day research — an article, a funding round, a competitor move, a hiring signal. Nothing is pre-generated.
    2
    The model writes one post
    Hook, argument, conclusion and closing question. Self-corrected up to three passes when a draft is close but not shippable.
    3
    On-image wording chosen
    The model picks the creative angle and the words to paint. Exact casing and dimensions are code's job, not the model's.
    4
    Deterministic HTML renders
    Typography is rendered by code — 1200×1200 and 1200×627, auto-fit. AI paints only the background.
    5
    QA before send
    Image text, header, card, identity and render are audited. Every figure is validated against the factsheet before it can ship.
    6
    Approved, then emailed
    One recipient per email, requester in CC. Send log plus Sent Items verification. No repeated topic, no same-day collision.

    Directory & listing presence

    The Manifest50–249 listed
    Clutchstaging only
    leadgenerationcompanies.in5-star listing
    RocketReachlive
    Datanyzelive
    ZoomInfolive
    SignalHirelive
    Instahyrefeatured employer
    Glassdoorvia Instahyre
    Company blog99 articles
    What the model decides vs. what the code guarantees
    Hook, argument, conclusion
    Every figure appears in the factsheet
    The creative angle
    Exact Title Case casing
    Which on-image words to use
    1200×1200 / 1200×627, auto-fit
    —
    One recipient per email, requester in CC
    —
    No repeated topic, no same-day collision
    Our point of view

    Eight things we can see that a list vendor cannot.

    Each of these is a property of the combination — behavioural capture, identity resolution, licensed intent and delivery all on one stack. Remove any one and the point of view narrows.

    01
    Behaviour, not just visits
    We know the pricing page was opened three times in a week, that the cursor stopped on the enterprise tier, and that the PDF was open for six minutes. Page-level intent proof — the top grade — not a modelled score.
    Grade A proof
    02
    Anonymous resolves to named
    An IP-matched email event and a site visitor are reconciled into one prospect record inside a 90-day match window — with device, OS, browser and organisation carried across. The trail never resets when a person changes role, device or email address.
    Identity resolution
    03
    Intent with a source and a date
    Every signal names where it came from and how old it is. Twelve-month-old interest is not a lead. Surge watch flags the topics accelerating right now and tells you the day and window to reach out.
    Freshness windows
    04
    The whole buying committee
    Breadth across the committee is part of the score, and roles are deduplicated rather than counted twice. A team reading in parallel looks different from one person reading alone.
    Committee breadth
    05
    Deletion and suppression are structural
    DNC, disposable domains and role mailboxes are blocked before an agent can pick a record up. Consent is captured at the point of download and applied to every downstream channel.
    30 / 60 / 90-day windows
    06
    Self-hosted measurement
    The tracker is our own plugin and our own dashboard. Visitor data is not handed to a third party beyond the free IP lookups, and the export only ever reads.
    No third-party handoff
    07
    The stack is read, not guessed
    Twelve categories with named vendors, an evidence trail and a confidence score. A peer set is estimated per archetype and the competitive overlap is scored — so the pitch matches what they actually run.
    144 named vendors
    08
    A human closes it
    The final step is a booked meeting with a verified decision maker, not a form fill. The pipeline ends where revenue starts — that is the difference between a data vendor and a demand partner.
    Human-verified revenue
    Market position

    The industry grades intent on evidence. We are graded A.

    Proof grade — whether you can see the exact page, timestamp and account, or only a modelled stage — is the honest way to compare these vendors. It is also why the same answer costs us less than one contract from an enterprise intent platform.

    A Page-level — exact page, content, URL, timestamp and account B Topic-level — account + topic + score, but no exact page C Modelled — a predicted stage, underlying evidence opaque
    VendorTypical costIntent proofWhat you actually see
    Mandit GlobalFree $250 ScaleAExact page, timestamp, account, source, freshness window
    BuiltWith$295–$6,000/moBTech / stack changes, no account-level intent
    Wappalyzer$250–$850+/moBStack-change alerts by domain
    HG Insights$12k–$90k+/yrBInstall events, topic-level intent
    ZoomInfo~$15k–$40k+/yrCStreaming intent, modelled stage
    Cognism~$16.5k+ (median ~$33k)BBombora topics, no exact page
    6sense$50k–$300k+/yrCPredicted buying stage, opaque evidence
    Demandbase$24k–$164kBAccount + topic + score, no page
    Bombora$25k–$300k+/yrBTopic + Surge score, no page
    Clay$185–$495/moAAggregates intent sources, traceability
    PredictLeads$0.04/credit, taperingAJobs, news, funding — source-linked
    Tool plans — every limit compared
    Free
    $0
    3 / day
    Pro
    $40
    25 / day
    Growth
    $109
    75 / day
    Scale
    $250
    300 / day
    Stack categories 12 Keyword intelligence Growth Buying intent Scale Source attribution Scale Surge webhooks Scale Daily allowance shared Yes

    Channel performance

    Conversations per 1,000 accounts reached · last 12 months
    Intent email84
    Syndication71
    Webinar63
    ABM outreach58
    Retargeting46
    Timing, hours & holidays

    We don't send when we are ready. We send when they are.

    Every send is placed inside the prospect's own working hours, on a day that is not their public holiday, in their own timezone — read from live data, not a spreadsheet we updated once.

    Working hours by region
    Local clock. Best window is where a reply is most likely.
    6 regions
    Best open time10:00 local — the model default
    Weekend weight0.22× — a Saturday send is a fifth of a Monday
    Peak10:00–11:00 at 1.18×
    Hour-of-day delivery curve
    Relative delivery weight per hour — click any hour to inspect it.
    24 hours
    Public holiday calendar
    50 countries · 307 entries · 3-day lookahead
    Live
    Send windows
    Each wave is bound to a named timezone, not a generic offset.
    8 waves
    Small factors

    The 5-point details that decide the outcome.

    A good list still fails if the domain was registered last Tuesday. These are the small factors we score, and none of them are decorative — each one moves the number, the status or the send.

    01 Audience temperature
    Warmth is not a vibe. It is a measured shift in open rate, click rate and spam complaint rate — and it compounds with the list source.
    02 Where they came from
    A purchased list is not a smaller version of an organic one. It is a different campaign with different rules.
    03 Lead-level detail
    Every check writes a note back onto the lead, so a human can audit exactly why a record was held back.
    The short version

    Most vendors sell you a list.
    We sell you the reasoning behind it.

    Behaviour captured first-party. Intent licensed and dated. Identity resolved to one account. Every record scored in the open. Every send consent-gated and suppressed. One outcome: a booked meeting with a verified decision maker.

    Founded 2018 · Houston, Texas & Pune, India · American General Center, 2929 Allen Parkway, Suite 200, Houston, TX 77019
    +1 254 277 8874 · contact@manditglobal.com
    navigate · O overview · F fullscreen
    Slide overview