Viktor.comAn AI coworker that actually does the work
Othor AI is an advanced business intelligence platform built on the best practices of the world's elite intelligence agencies and 100+ years of signal analysis theory. Othor AI is the AI-native, fast, easy, and secure alternative to popular BI tools like Power BI, Tableau, and Looker - built from first principles for the way businesses actually work. 100x better than using Claude or ChatGPT alone and at a fraction of the cost of your existing LLM & BI stack.
This is the 2nd launch from Othor AI. View more
Othor AI 2.0
Launching today
BI assumes you have an analyst, a warehouse, and time. Leaders have none. Othor inverts it: connect a source, and agents start analyzing on their own.
10 new features, including Data Integration Agents (100+ sources, no migration), Analytical Agents, Decision Briefs, daily Big Picture reports, Document Intelligence, Chat with Data, Dashboards, and a Model Selector for your own LLMs.
They compose - sources feed agents, agents feed briefs, briefs feed Big Picture. One pipeline, three altitudes.
Free Options
Launch Team
Unni covered why we started, so I'll take the other half. Every time I read an AI product post my first thought is "okay, but is it a chatbot with a nice landing page," and that's a fair thing to think about us too. So rather than tell you it works, let me walk through the four decisions we made, including the ones that made the build harder.
1. The LLM doesn't calculate anything
Ask a language model to do arithmetic on your business data, and you get an answer that sounds right. Sometimes it is. That's fine for a blog post and not fine for someone deciding whether to pull a product from a region.
So we split the work. Our analytics engine computes every number deterministically, in SQL, against your actual data. The model figures out what's worth looking at and then explains the result in plain English. Numbers from the engine, words from the model. An analytics engine with an LLM on top, not an LLM with a chat box.
Nice side effect: our accuracy doesn't move around based on whichever model everyone's excited about this month.
There's a "Calculation" button on everything. If you can't check it, you shouldn't trust it, and that includes checking us.
2. No warehouse, no migration, no six month project
Traditional BI wants your data moved somewhere before it'll do anything. That's a quarter of work before a single person sees a chart, and it's why so many BI rollouts quietly die before they deliver.
We connect live to whatever you already have. Point Othor at a database and it reads the schema, samples the data, and the agents start working on their own. Postgres, MySQL, SQL Server, Snowflake, BigQuery, Mongo, S3, Sheets.
Your raw data stays exactly where it is. We hold connection details, schema, metric definitions, and the aggregated results. Not your rows.
3. Nobody should have to ask the first question
This is the bit self-service BI never solved. It hands you a search box and assumes you know what to type. Most people don't, and not because they're not smart. They just don't know what's sitting in the data. You can't ask about a thing you've never seen.
So Othor goes first. The agents work out which metrics actually matter for your business, compute them, rank them by whether the movement is genuinely interesting or just noise, and write up what changed and what probably caused it.
Every morning there's a Decision Brief waiting, plus a Big Picture view that pulls the whole workspace together. You walk in already knowing what moved overnight. After that, sure, chat with it, drill in, build a dashboard if you like dashboards. But the first insight shows up without anyone asking.
4. Your data can't tell you why
Every BI tool is staring at the same thing. Your data. But most of what actually moves a business starts somewhere outside it.
Yarn goes up 18%, and your material cost follows next month
The rupee moves, and your import margin quietly changes shape
Panel prices shift, and the product you built your margin on is no longer the product you built your margin on
None of that lives in your database. By the time it turns up in your numbers, it's already happened to you.
So we look outside as well. Signal Discovery Agents watch curated external streams: commodities, currencies, energy, shipping, weather, news. They pick out the moves that matter, find the article that explains why it moved, and tie it back to the internal metric it's likely to hit. It shows up as a banner on the metric itself, not in some separate feed you'd open twice and then forget about. Yarn up 18%: here's the story, here's what it might do to your material cost.
That's roughly the difference between a tool that tells you what happened and one that tells you what's coming.
What's actually in 2.0
Data Integration Agents. Connecting a source takes a minute, not a migration
Analytical Agents. Find the metrics, do the maths, write the narrative
Decision Briefs and Big Picture. The daily read on what moved and what it means
Signal Discovery Agents. The external stuff above
Document Intelligence. Because a real answer usually needs the contract as well as the spreadsheet
Chat with Data. For when you do have a specific question
Dashboards. For when you want the classic thing
Model Selector. Bring your own LLM, self-host it, or run open source
We're model agnostic on purpose, and we'll run on cloud or fully on-prem if your data can't leave the building.
One thing I'd like to be argued with about
We bet that proactive beats conversational. That a brief sitting there at 6am is worth more than being able to ask a brilliant question at 3pm. Almost everyone else is betting on the chat box.
If you think we've called that wrong, I actually want to hear it, and I'll be in the comments all day.
Try it, break it, tell me where it falls over. Far more useful to us than an upvote.
Hey Everyone! I’m Nekender, co-founder and CTO of Othor AI.
Unni has already talked about the vision. I want to talk about why we refused to ship another “chat with your data” box, and what that forced us to build.
The problem we kept hitting
Every serious company I have worked with already has data. What they do not have is a trustworthy picture of the business that shows up in time to act.
The usual path is still connect a warehouse, hire a data team, model a semantic layer, wait months, then maybe ask a chatbot questions on top. If you skip that and just point an LLM at raw tables, it invents metrics that look perfect and are wrong. Executives try it once, lose trust, and go back to Excel.
Othor AI is not the chat window, Othor AI is the living metric system underneath it.
What Othor actually does
You connect operational systems and files, databases, spreadsheets, the messy stuff a data team was supposed to clean later. Othor then:
Discovers what is actually measurable in that data, and throws away what is not trustworthy.
Builds metrics that can span sources that were never designed to join, without forcing a lakehouse migration first.
Treats units and currencies as a first-class problem, so a number from two systems is actually comparable.
Proposes which external market and economic series appear to move your internal KPIs then lets you confirm or dismiss.
Runs always-on agents against goals. They produce narratives, decision briefs, and a daily Big Picture. You do not have to know what to ask.
Chat is there. Charts are there. They are the front door. The hard part is everything behind them.
What we learned the hard way
Language models are excellent at sounding sure. They are terrible at inventing KPIs that survive a CFO. Most of our engineering time has gone into that gap making sure a metric is computable, current, unit correct, and not a hallucination with a pretty name.
The other lesson companies do not live in one warehouse. They live in five systems and three Excel trackers. If your product needs a clean semantic model before it can help, you have already lost the people who need you most.
And internal numbers alone are not enough. A persistency drop or a margin swing often starts outside the company. If your BI cannot talk about the outside world with any discipline, it can tell you that something moved and never why.
What I would love feedback on
First connect experience. We want “useful in minutes,” not “useful after a data project.”
Trust. If a number feels off, tell me. That is the product.
Agents and briefs. Do they save a Monday morning, or do they still read like a dashboard in paragraph form?
External signals. Which outside world series would you actually want watched against your business?
I built a lot of this stack myself backend, product surfaces, and the unglamorous work that keeps an insight from being confidently wrong. If you are a founder, operator, or data person who has been burned by pretty AI analytics, I want the honest version of that story.
Try it at othor.ai, share your feedbacks in the comments.