Dayonik is an AI investing platform that pairs a mean-variance optimisation engine with an adviser which only ever voices what the maths has already worked out. There aren't any generic guesses here. You get findings you can audit, targets you can act on and the tax you'll owe shown plainly before you trade.
The optimiser sits right at the centre, and it is the part that everything else depends on. The advisor, the scheduled reviews and the tax logic are all built on top of it. So whatever the platform tells you, it is grounded in the very same quantitative core.
Deterministic findings, and every one of them is tied to a computed number such as concentration, volatility, drawdown or the distance to the frontier. A language model voices the advice, yet it never gets to decide it.
Mean-variance, max-Sharpe, risk-parity, target-return and more. You set per-asset bounds, group caps and an efficient frontier, whilst the trade generation stays aware of transaction costs.
A per-lot ledger sorts every sale into short-term or long-term. What is more, the rebalancer leans towards min-tax lots and it shows the estimated tax cost right next to the commission. In other words, you optimise after tax and not merely before it.
Save an investment policy, bind it to a portfolio and let Dayonik monitor it for you. Scheduled reviews and drift alerts reach you at the very moment a rebalance is worth doing.
"Your portfolio looks a bit risky and could be more diversified. Consider trimming your tech exposure and adding some bonds for stability."
It is plausible, and impossible to verify. It comes out different every time you ask. There is nothing at all to audit.
"NVDA is 80% of the book, above the 35% guardrail. Rebalancing to the max-Sharpe target lifts expected Sharpe 0.41 to 0.90."
Every scheduled review reduces to a short, ranked set of findings. Each one carries the evidence it rests on, together with the exact trades that bring the book back to target. And it only lands in your inbox when something genuinely matters.
NVDA is 80% of the portfolio, well above the 35% guardrail, so one single name drives an outsized share of your risk.
Expected Sharpe could move from 0.41 to 0.90, which is more return for every unit of risk.
Import your holdings or log your trades. From the very first transaction, Dayonik keeps a per-lot cost basis, and that is exactly what makes all the tax-aware features work.
Choose a risk profile, or build your own objective and constraints. Then save it as a strategy you can reuse and bind it to a portfolio.
Now leave it running. Scheduled reviews and drift alerts surface a grounded recommendation the moment a rebalance is worth the tax and the cost.
Dayonik is a new AI investing platform, so we are not going to invent a track record we have not earned yet. What we will do is state plainly how the product behaves, and every one of these is something you can check for yourself the moment you sign in.
Every finding is tied to a computed number you can see. The model voices the advice. It never gets to decide it, and nothing reaches you that the maths has not worked out first.
Run it hosted or self-host on your own Postgres. Either way the audit trail is there in full, so you can trace any recommendation back to the numbers behind it.
It is free to start and there is no card required. You upgrade only when your book does.