The loop, end to end
How Solon works
You write down why you own something and what would make you wrong. Solon reads against it every night, hands your own plan back the moment it breaks, and scores the call you made. Below is that loop, shown as what you actually see.
Ink is yours. Azure is drafted. You always know whose reasoning you are reading.
You say why you own it. Solon makes it testable.
Talk for a minute, typed or spoken. Solon turns what you said into named claims, each carrying a number a filing can settle, plus the thing you will do if it breaks. You approve every line before it counts.
“Enterprise workflow spend keeps consolidating onto ServiceNow. If contracted growth slips under twelve percent two quarters running, I want out of half.”
If it breaks: exit half. (your precommitment, kept for later)
Every night, Solon grades your claims against what actually happened.
Not sentiment, not a price alert: every filing, headline, print and rival launch is asked one question, does this press something you committed to in writing. The model reads and reports facts; the grade is computed by rules in code, so the same finding always scores the same way.
The filings agent. Q3 10-Q landed at 4:31 pm. Contracted growth printed 14.2%, above your 12% floor. Claim settled from the document.
The news grader. Rival guidance cut in the wires: touches nothing you wrote. Left alone.
The ecosystem agent. OpenAI shipped enterprise agent workflows. You never listed them; they press your franchise claim, so they joined your roster.new rival
The earnings agent. Q3 prints in 6 days. It tests your 12% floor directly.
The verdict engine. Recomputed both claims from your own rules on the close. Still on track, and the franchise claim is the one to watch into Thursday.
When a claim breaks, Solon reads your own plan back.
No opinion on the stock, ever. What Solon has is the sentence you wrote while calm, and the fact that its condition just happened. It says both, asks one question, and dates your answer. Never wrote a plan? It says so rather than inventing one.
Contracted revenue growth printed 11.4%, the second consecutive quarter under your floor.
You said: “If it breaks: exit half.”
Are you holding through this?
Solon scores the call, three ways, kept apart.
What the market did after. Whether you kept the rule you set before it moved. And your own reasoning, quoted back. Never merged, because a well-reasoned call that lost and a lucky call that broke your own rule must not read the same. Entry calls land in the same ledger, including the zones you looked at and passed on, and Solon grades its own calls separately on the same terms.
Your own floor was $620, and the price was $607 when you were asked. You held anyway.
“Agent backlog is real. One soft quarter does not kill the franchise.”
Would you make it again? Asked at your next sit-down.
Several positions can be one bet. Solon tracks the bet.
Five tickers can rest on one belief. A view groups them, puts the claim underneath the whole group, and when the argument weakens it weakens across every position resting on it.
One claim underneath: enterprises keep buying agentic software from the incumbents that already hold their workflows.
What Solon does
Everything below runs today. One product, one price: none of it is a tier you reach later.
The positions you hold
- Read-only brokerage link or a positions file; cost basis comes with it
- Your reasons become named claims, spoken or typed
- A nightly verdict per position, from your rules, not sentiment
- The street's side: analyst spread and its shift, the multiple as a percentile of the name's own history, the 52-week position
- What each company is judged on, learned from its own release
Exposure across the whole portfolio
- A map of every holding, sized by what it is worth to you
- Views: one bet spread over several tickers, charted against a benchmark
- Themes computed from what you hold, with the weight on each
- Drift between the size you intended and what it became
- Holdings with no thesis are named, never silently counted
The names you are still deciding about
- A watchlist where each name gets a drafted thesis before you own it
- An entry zone, the condition worth waiting for, and what entering looks like
- You are told when a price reaches a zone you drew
- Passes are scored like buys: the zones you declined are part of the record
Research, against your own positions
- Describe a belief; get names that fit it, reasoning shown
- An AI chat that reads your own exposure, weights, rivals, record, news and quotes, searches the web, and proposes edits you approve
- A plain explanation of what each company you own actually does
- Rivals you never listed, found and graded against your thesis
The record of how you decide
- Every decision dated, with your words and what the market did after
- Whether you kept your own floor, tracked apart from whether it paid
- Post-mortems on closed positions, measured since your exit
- Solon's own calls graded in a separate ledger, same terms
- A weekly sit-down that hands back the calls old enough to judge
What reaches you, and when
- One digest, only on days something actually happened
- An outsized intraday move is sent when it happens
- Before each print, what that quarter tests in your case
Investment products model securities. Solon models the investor.
The decision is the object, not the alert. The alert is one step in a record of what you committed to, whether you followed it, and what that was worth.
The record becomes the system of record for your judgment. Which breaks you trade through, whether you sell too early, whether your entry zones are priced well. No market-data product holds that.
The numbers never come from the model. Every figure is read from a filing or a feed; the model writes prose around it. Enforced in code, not by prompt.
Your words are marked as yours. Ink for what you wrote, azure for what Solon drafted, on every surface.
The same process, without the desk
What each of those four roles does at a fund, and which of them you can run yourself, is on the hedge fund process, for one person.
The case is written down before the money moves
An analyst reads against it
A risk desk watches the levels
When it breaks, a post-mortem
What Solon will not do
Solon never executes a trade, never moves funds, and never holds custody of any asset. A brokerage link is read-only. Solon is not an investment adviser, issues no ratings and no recommendations, and every verdict is computed mechanically from claims and levels you wrote yourself. More in the FAQ.
Everything above assumes the claims exist. Choosing which ones are worth carrying is the harder half and it is set out on how to write an investment thesis. The practice this belongs to is investment thesis monitoring, and the failure it exists to catch is thesis drift.
Your thesis is already written. It is just not written down.
Say it once, and your agents read every night after.
Already invited? Create your account