Today, we're excited to welcome @triton_research to the Carbon Arc ecosystem - bringing structured visibility into the vast universe of private companies that operate outside the reach of traditional financial data. Spanning 120,000 private companies across 180 countries, Triton aggregates fragmented public signals into standardized profiles - delivering modeled financials, workforce intelligence, and funding activity, all normalized and joinable to Carbon Arc's broader data ontology. As capital flows increasingly chase private markets and the line between public and private blurs, this dataset gives users a rigorous view into private company valuations, peer benchmarking, sector momentum, and the emerging sub-industries shaping where growth is heading next. If you're sourcing private investments, benchmarking portfolio companies, or mapping thematic trends across private markets, we'd love to connect. Reach out here to learn more: carbonarc.ai/contact-us

The @lightpathtech CEO congratulating @LockheedMartin on their NGSRI website isn’t subtle but IS much appreciated 😉 All signs point to the fact that @LockheedMartin has likely won the NGSRI contract given this new website and recent NGSRI related hiring listings. If so, look out for a $LPTH rerating. If won, @lightpathtech will become the @LockheedMartin supplier of the advanced thermal camera systems used to support the Army’s NGSRI missile program. If won, $LPTH likely to generate an additional $50-100m in long term revenue. @BussinBiotech has a great chronology thread on the NGSRI timeline here: x.com/BussinBiotech/status/2…

Shein's acquisition of Everlane made headlines. @CarbonArcAI data tells the real story. Across the top 50 cross-shopped brands for each retailer, only one appears on both lists. @SHEIN_Official's customer base is 50% Gen Z, 53% from households under $45K. @Everlane's customer base skews older, with higher household incomes. Shein's customer base shops frequently, spends less. Everlane's shops deliberately and spends more. See the full data-driven rationale in the thread. @business @BoF @hm @ZARA @Reformation

Prediction markets for company KPIs is a game changer. This is the markets that pulls every Institutinal investor in. Think about markets where you can predict for $UBER the number of rides last quarter. Or the number of DoorDash deliveries. Or even the number of vehicles that $TSLA will deliver. Hedge funds, especially the pod shops are spending a ton of time and effort forecasting quarterly earnings. But the teams, especially data science teams, have always been good at using unique data to predict the KPIs. You can’t control if the stock goes up or down on earnings announcements. But now you can make money on being the most accurate to predict KPIs. Or even use prediction makets on KPIs as a hedge. This is the market that brings in billions of $$$ from institutional traders. It’s a no brainer. When I worked at a hedge fund we used credit card data, foot traffic, camera footage, satellites and endless other sources to predict KPIs. And we were very accurate and even more importantly we were great at being directionally accurate (beat or miss). Now this is a whole new market of opportunity to place your predictions on these KPIs! Data just got more important and more valuable, especially if it can predict KPIs! Go get the @fiscal_ai API so you can start predicting on @Kalshi

LLM Knowledge Bases Something I'm finding very useful recently: using LLMs to build personal knowledge bases for various topics of research interest. In this way, a large fraction of my recent token throughput is going less into manipulating code, and more into manipulating knowledge (stored as markdown and images). The latest LLMs are quite good at it. So: Data ingest: I index source documents (articles, papers, repos, datasets, images, etc.) into a raw/ directory, then I use an LLM to incrementally "compile" a wiki, which is just a collection of .md files in a directory structure. The wiki includes summaries of all the data in raw/, backlinks, and then it categorizes data into concepts, writes articles for them, and links them all. To convert web articles into .md files I like to use the Obsidian Web Clipper extension, and then I also use a hotkey to download all the related images to local so that my LLM can easily reference them. IDE: I use Obsidian as the IDE "frontend" where I can view the raw data, the the compiled wiki, and the derived visualizations. Important to note that the LLM writes and maintains all of the data of the wiki, I rarely touch it directly. I've played with a few Obsidian plugins to render and view data in other ways (e.g. Marp for slides). Q&A: Where things get interesting is that once your wiki is big enough (e.g. mine on some recent research is ~100 articles and ~400K words), you can ask your LLM agent all kinds of complex questions against the wiki, and it will go off, research the answers, etc. I thought I had to reach for fancy RAG, but the LLM has been pretty good about auto-maintaining index files and brief summaries of all the documents and it reads all the important related data fairly easily at this ~small scale. Output: Instead of getting answers in text/terminal, I like to have it render markdown files for me, or slide shows (Marp format), or matplotlib images, all of which I then view again in Obsidian. You can imagine many other visual output formats depending on the query. Often, I end up "filing" the outputs back into the wiki to enhance it for further queries. So my own explorations and queries always "add up" in the knowledge base. Linting: I've run some LLM "health checks" over the wiki to e.g. find inconsistent data, impute missing data (with web searchers), find interesting connections for new article candidates, etc., to incrementally clean up the wiki and enhance its overall data integrity. The LLMs are quite good at suggesting further questions to ask and look into. Extra tools: I find myself developing additional tools to process the data, e.g. I vibe coded a small and naive search engine over the wiki, which I both use directly (in a web ui), but more often I want to hand it off to an LLM via CLI as a tool for larger queries. Further explorations: As the repo grows, the natural desire is to also think about synthetic data generation finetuning to have your LLM "know" the data in its weights instead of just context windows. TLDR: raw data from a given number of sources is collected, then compiled by an LLM into a .md wiki, then operated on by various CLIs by the LLM to do Q&A and to incrementally enhance the wiki, and all of it viewable in Obsidian. You rarely ever write or edit the wiki manually, it's the domain of the LLM. I think there is room here for an incredible new product instead of a hacky collection of scripts.

Ahead of $NKE (Nike) earnings on 3/31. See how Carbon Arc Lenses, an AI powered research interface on top of our full external data catalog, models Nike's last quarter and surfaces what to listen for during the call. Watch the video to see how quickly it is to get an earnings read. Request the full report below.

Using @CarbonArcAI, I pulled transaction and web traffic data for Quince. You can see the tremendous growth since 2023, peaking at over 1,400% growth in credit card spend over the holiday season!

This is super interesting as well I've been experimenting with AI driven data dashboards to more systematically track company KPIs. So much of the public-equity research process is about developing better revenue forecasts (which then flow down to EPS at various incremental levels). Thus, so much of the investment research motion is about tracking data that informs more accurate revenue forecasts. This is the foundation of the alternative data industry. But there are many helpful data sets sitting in the open. As chatbots have grown arms (Claude), I have been impressed by the ability of these tools to go grab this data. x.com/FundamentEdge/status/2… To me, this is the other missing piece that sits alongside Excel fluency. The ability to ingest an Excel model (eventually build it, but not today), identify key drivers (from open and proprietary data), distill that data back into forecasts, and flag back custom alerts (business momentum inflections, likely revisions to revenue forecasts, thesis validation/invalidation, etc). Now we are talking!! This is orders of magnitude more helpful than a finance chatbot wrapper. I do wonder if these tools are progressing at such a pace that most finance professionals don't even need to learn coding agents. For example, Perplexity Computer one-shotted something that, for sure wasn't perfect, but exhibits material progress in the capability of the harness infrastructure to build simple, powerful user interfaces. Institutional grade accuracy remains a critical & still not fully solved issue (does this improve as firms like CarbonArc roll out MCPs??), but it is exciting to see the engineering capabilities improve so materially, in such a short period of time.

If you have the Carbon Arc MCP Server connected in @claudeai, you have it in Excel! Seamlessly ask for and bring data from Carbon Arc right into your spreadsheet. Demo below 👇