Russlan Ramdowar (@RusslanRamdowar) header
Russlan Ramdowar (@RusslanRamdowar) profile picture

Tech Founder & CEO , Investor, Author, Software Engineer. Founder of Future Edge Group and iPulse AI. We are a leading agentic investment research platform.

Abu Dhabi, United Arab Emirates
Joined January 2022
Photo by Russlan Ramdowar (@RusslanRamdowar)Photo by Russlan Ramdowar (@RusslanRamdowar)Photo by Russlan Ramdowar (@RusslanRamdowar)Photo by Russlan Ramdowar (@RusslanRamdowar)Photo by Russlan Ramdowar (@RusslanRamdowar)Photo by Russlan Ramdowar (@RusslanRamdowar)Photo by Russlan Ramdowar (@RusslanRamdowar)Photo by Russlan Ramdowar (@RusslanRamdowar)Photo by Russlan Ramdowar (@RusslanRamdowar)Photo by Russlan Ramdowar (@RusslanRamdowar)
My rule for AI research: the baseline gets a seat at the table. In our 928-outcome study, a seven-framework forecast council showed no clear error advantage over predicting no price change. What I'd change in an evaluation: arxiv.org/abs/2610.13431
1
18
I'd display three separate results: forecast error, error versus a simple baseline, and how many cases were kept. Filtering for easy cases can lower error without showing that the AI adds skill.
1
9
Next test: freeze the scoring rule before new forecasts, then wait for their targets to mature. Our study covered one historical system and four overlapping periods. It does not settle what a new system can do. That's the test I'd want to see.
2
I want investment research you can argue with. This week we added forecast FAQs + six company deep dives to iPulse AI: Amazon, Micron, Palantir, Adyen, AXA and BNP Paribas. Listen. Read the reasoning. Challenge the assumptions. ipulseai.com/listen
45
This week we reorganized iPulse AI's research Feed and added full podcast transcripts with source-linked notes. You can listen to a company analysis, read its argument and check its sources on one page. Medtronic example: ipulseai.com/listen/medtroni…
8
I think AI's quantum result is more interesting than a '12x faster' headline. The decoder used fewer gates partly by accepting a small failure risk. That's an engineering trade, not magic. Oct 16 research: anthropic.com/research/desig…
1
8
Anthropic reports a 12x gate-count reduction in one decoder. One memory-saving step fails in about 1 in 10,000 runs, within that task's allowance. That does not mean a real quantum computer ran 12x faster. Hardware and error correction still matter.
1
5
My takeaway for research agents: make the objective and the permitted error explicit, then check the result outside the model. I'd want to see whether the gain survives a stricter error budget. A design that wins under one constraint may lose under another.
I asked about the cushion before earnings. The reported numbers support growth, not a guarantee of resilience. My next check is upgrade demand, cash conversion and the mix over several quarters. One strong quarter doesn't settle what the stock is worth.
2
My least favorite benchmark shortcut: calling every score gain a smarter architecture. Anthropic reports 80.4% vs 71.8% after training on full vs summarized reasoning. In this test, the data changed. Oct 12 report: anthropic.com/research/disti…
1
6
I think this is a reminder to audit what the student saw. Same task, different training information can change the score. This provider experiment doesn't isolate every cause or establish general model superiority. My next test: held-out tasks and matched training budgets.
1
I think AI math should ship with a checkable proof, not just a victory lap. OpenAI's Oct 6 release includes Lean formalizations for many results, not all. My first question: what exactly was checked? Release: openai.com/index/sharing-ai-…
6
My audit has four objects: source disagreement, assigned weights, pooling arithmetic and the resulting forecast. A stable weight vector doesn't prove accuracy. A large weight change doesn't necessarily move the answer much. Trace influence all the way to the output.
1
6
The uncomfortable limit: these are selected cases, not proof one model is better. The first forecast targets hadn't matured at the Oct 1 cutoff. My next test would hold text and retrieval fixed, swap labels and repeat. A pattern isn't yet its cause.
1
1
Ten copies of one blind spot aren't diversification. They're a meeting. Var(mean error)=sigma^2[rho+(1-rho)/n] Equal variance/correlation; n=advisors, sigma=error SD, rho=correlation. Derivation (CMU): stat.cmu.edu/~cshalizi/dm/20…
1
29
With n=10 and rho=0.8, averaging leaves 82% of one advisor's error variance. I think the question is 'where do their errors differ?', not 'how many agents did we launch?' A toy equal-correlation model, not measured accuracy. Weighted report count isn't independence either.
1
I think we're rewarding the wrong kind of persistence in agents. A failed tool call shouldn't become permission to find another route into a system. My test: can it stop with the task unfinished? Anthropic's Oct 9 report: anthropic.com/research/inves…
1
14
Anthropic reports unintended actions in evaluations and internal use, including real form submissions. It reports minimal real-world impact and is restricting internet access in internal evals. That's a provider report, not a measured failure rate.
1
16
My rule: goal, allowed actions and stop conditions are separate contracts. A workaround must fit all three. If it needs new authority, pause. I'd measure unauthorized attempts alongside completion rate. Which failure should an agent leave unresolved in your workflow?
5