SkymelGet started

Mine your own intelligence
from the open web.

Mel mines the web’s smallest signals to build a knowledge graph around your goal. Then ask Mel to use that graph to create custom intelligence reports, build datasets or monitor important changes, for you and your agents.

interviewed bybriefedattendedspoke atpostedvisitedmetspoke atattendedsignedreceivedmetspoke atbriefedLee Jae-myung3 of themWe Must Pace the Fronti…4 of themFortune 500: Titans and…14 Sept 2026Dreamforce 20263 of themUN Security Council2 of themTrump-Xi state dinner4 of themWhite House AI accord4 of themAI and religious author…3 Oct 2026Elon Musk16 Oct 2026Navier-Stokes milestone…16 Oct 2026SEGA 30th anniversary,…15 Jul 2026YC Startup School 202631 Jul 2026Hugging Face3 Sept 2026National Medal of Scien…2 of themAnderson Cooper 36015 Sept 2026Copilot as a new OS25 Sept 2026Windows hybrid intellig…7 Oct 2026National Medal of Techn…8 Oct 2026AI emergency brake10 Oct 2026Chair of Google DeepMind5 Aug 2026Gemini 4 Argon wizard c…13 Oct 2026Building a laser with A…13 Oct 2026International AI Summit…14 Oct 2026Wall Street Journal21 Oct 2026Meta Q2 earnings call29 Jul 2026Meta Connect 202623 Sept 2026Cheeky Pint13 Oct 2026Cheeky Pint Guinness le…13 Oct 2026Guinness grimace clip13 Oct 2026AMD Advancing AI 202623 Jul 2026CNBC Squawk on the Stre…5 Aug 2026World Labs28 Sept 2026Taipei suppliers and TS…5 Oct 2026OCP Global Summit 202612 Oct 2026Inkling15 Jul 2026Lilian Weng28 Jul 2026Accel4 Sept 2026Weights people own15 Oct 2026Firmus21 Oct 2026Sam AltmanOpenAIJensen HuangNVIDIADario AmodeiAnthropicDemis HassabisGoogle DeepMindMira MuratiThinking MachinesSatya NadellaMicrosoftMark ZuckerbergMetaLisa SuAMD

What teams have uncovered with Mel.

We used Skymel to identify high-potential school photography prospects in France and Belgium. It went beyond just building a prospect list. It researched each company in depth, identified relevant signals, and helped us understand why and when a prospect might be worth pursuing. We particularly liked how the research translated into clear recommendations on who to approach, why they were relevant, and what our sales angle should be. It saved our team significant research time while producing insights that would have been difficult to build manually at this scale.
Eric SémanneCEO, PixFirst
We gave Mel a single objective. What came back was far beyond what we expected. A structured view of the AI PR landscape, the agencies doing the most relevant work, the clients behind that work, and the narratives connecting it all. We were able to turn that intelligence directly into a focused go-to-market strategy for PromptPR.
Wilson CraigSVP/GM, MindsharePR · Co-Founder & CEO, PromptPR

Billions of small activities. Traced across millions of websites, every day.

Mel finds the few thousand microsignals relevant to your goal and converts them into a knowledge graph built around your goal.

LinkedInJob postsXPublic filingsRedditResearch papersYouTubePodcastsGitHubConference talksTikTokProcurementInstagramPatent filingsThreadsRegulatorsFacebookTrade pressDiscordCompany sitesStack OverflowRelease notesHacker NewsEarnings callsMediumStatus pagesSubstack

Go beyond what’s easy to find.

The detail you need can be anywhere. Mel finds what matters to your goal, connects it, and creates a knowledge graph relevant to your goal.

Step 1

Tell Mel what you want to mine from the web.

You tell Mel exactly what you want to find on the open web. It takes your goal and sets out to mine the relevant activity.

Example: Where is physical AI becoming real business?

LinkedInYouTubeRedditGitHubJob postsPapersNewsNewslettersCompany sitesX+ millionsmore websites, every day
Step 2

Uncover the smallest signals that matter.

Mel searches the web for microsignals related to your goal, checking sources ranging from forum replies to footnotes in filings. It collects these small details to connect them into a larger picture.

Example: Where is physical AI becoming real business?
youtube.comOne Company Deployed 1 Million Warehouse Robots
github.commicrosoft/physical-ai-toolchain
jobmesh.ioDepot Specialist (Robotaxi Fleet Operations)
globemarketresearch.comUS Physical AI Market Size to hit USD 29.6 Bn by 2035
arxiv.orgTowards Trustworthy Physical AI: From Theory to Practice
substack.comPhysical AI: The Full Investible Stack
github.comuhub/awesome-python: A curated list of awesome …
robotics247.comHumanoid Orders vs. Actual Deployments: How Much Commercial Progress Is Really Happening
Step 3

Connect who, what, where and when.

Mel determines the specific companies and people each microsignal refers to. It links what happened with who said what across different names and web sources.

Example: Where is physical AI becoming real business?
LinkedIn post · 23 Jul 2026

BMW's Spartanburg, South Carolina plant tells a more precise story. Two Figure 02 robots spent 11 months on one task: inserting sheet-metal parts for the welding process, physically demanding and highly repetitive work. No workers were laid off. Those previously on the task were redeployed elsewhere in the facility.

RobotFigure 02CompanyBMWPlacePlant SpartanburgDuration11 months on one taskTaskSheet-metal parts for weldingClaimNo workers laid off
Step 4

Get a 360° view built around your goal.

Those relationships become a knowledge graph built around your original goal. This brings scattered details together into a connected representation of the real world over time.

Example: Where is physical AI becoming real business?
Step 5

Put your graph to work.

Intelligence reports
Mel investigates what causes a development by correlating signals and making inferences from the graph.
Custom datasets
Mel constructs the exact dataset you need from the entities and relationships it found.
Monitoring
It detects the smallest relevant changes over time and works out how they affect your goal.
Example: Where is physical AI becoming real business?

Carmakers are turning into humanoid-robot makers and buyers.

  1. Elon MuskTesla
    Tesla

    Tesla is ending Model S and X production to make space for Optimus manufacturing.

    Elon Muskstatement:claimModel Scnevpost.com
  2. Chung Eui-sunHyundai
    Hyundai

    Hyundai is planning a dedicated Atlas plant targeting 30,000 robots a year by 2028.

    Boston Dynamicsstatement:predictionHyundaicheatsheets.davidveksler.com
  3. Wang ChuanfuBYD
    BYD

    BYD has committed to deploying 20,000 humanoids in its EV plants.

    BYDrelatedToChinawisdomtree.com
  4. He XiaopengXPeng
    XPeng

    XPeng showed its humanoids at Hannover Messe in April 2026.

    Hannover Messe 2026event:participantXPengemag.directindustry.com
Mel infers

Carmakers already have the motors, batteries, AI chips and assembly lines humanoids need, so the industrial base for humanoids is forming inside car companies.

Points the other way

None of these plans has reached volume yet. Hyundai's is 2028, and Tesla says its units are still for learning.

physicalaijournal.org

The world doesn’t stop. Neither does Mel.

Keep Mel on it. Mel keeps your graph current, detects even the smallest relevant changes and explains what they mean for your goal.

relatedTorelatedTorelatedToevent:participantevent:participantAmazonAug 2026GXO LogisticsAug 2026SSchaefflerAug 2026ARAgility Robotics comm…Sept 2026D5Digit 5 UnveilingSept 2026Agility Robotics
Example: Agility Robotics
Mel’s read, stopped before Sept 2026Agility Robotics is a niche provider of humanoids for specific logistics testing within a handful of large corporate partnerships.highways.today

Agility Robotics reported over $300 million in multi-year commercial orders for its new Digit 5 humanoid.

faydcapitalgroup.com
Mel’s read, kept runningAgility Robotics has successfully transitioned from experimental pilot programs to a scaled commercial business with significant, multi-year revenue commitments.

One intelligence foundation. Many possibilities.

The same connected knowledge can answer complex questions, form specialized datasets, or power applications, for you and your agents.

Investigations

Go deeper into a question, market, relationship, or developing situation. Get connected findings, useful context, and clear conclusions.

Read the physical AI report
Intelligence report

Physical AI: beyond the robot.

  1. Customer tasks driving demand
  2. Suppliers across robot makers
  3. Operations and services
  4. Connected findings and sources

Custom datasets

Build structured intelligence around the entities, attributes, conditions, and relationships your goal requires.

Download the dataset
CompanyRobotWhat happenedBasis
BMWFigure 0230,000+ X3s supportedCustomer-disclosed
BMWFigure 03Arrived at Hall 52, 30 Jun 2026Company-claimed
TeslaOptimusInternal training units onlyYahoo Finance

Intelligence feeds

Deliver relevant signals, relationships, and changes into applications, internal systems, and AI agents.

Download the watchlist
{
  "entity": "Figure 03",
  "event": "arrived at BMW Hall 52",
  "date": "2026-06-30",
  "relationships": ["BMW", "Figure AI"],
  "source": "physicalaijournal.org"
}

Anything you want from the open web.

Your own goal

Your goal. A world of intelligence.

Start with a deep investigation. Build a custom intelligence resource. Or keep Mel working as the world changes.

From $150 per investigation · No integration required

Before you start.

What counts as a signal?

Any relevant public detail or activity: a sentence in a filing, a skill in a job post, a comment, a research footnote, an event appearance or a change to a public source. Its value often becomes clear when it is connected to other information.

Do I need to build or explore the graph myself?

No. Mel finds relevant information, resolves entities and connects the context around your objective. Ask for the intelligence report, dataset or monitoring you need; the graph supports the work.

Can I ask for a dataset instead of a report?

Yes. Define the entities, fields, coverage and output you need. Custom datasets are one of the ways to use the connected information Mel collects.

Can Mel keep monitoring after the initial work?

You can scope ongoing monitoring around your objective, so new activity and important changes can be understood using the context already collected.

How do I choose the scope and price?

Start with your objective, coverage and desired output. See Skymel’s current pricing for available options, or describe your requirements to scope the work.