How to Use the Thesis Drift Agent for Indian Stocks: A Practical Daily Workflow

What The Thesis Drift Agent Is Actually For

The Thesis Drift Agent is not trying to be a magical stock picker or an auto-trading engine. Its real job is much more useful: it watches your live watchlist, compares today's signals against the thesis you want to protect, and tells you which symbol actually deserves attention now.

That sounds simple, but it solves the hardest daily problem in public equities. Most investors do not fail because they have no information. They fail because they have too much disconnected information and no reliable triage system. The agent turns that chaos into a ranked work queue.

Start In The Live Dashboard

Open [Arthalekh Live](https://arthalekh.com/live). The Thesis Drift Agent sits inside the live dashboard beside your watchlist, live market pulse, and news signal modules. That placement matters because the right workflow starts with context, not with one isolated stock page.

If your watchlist is empty, fill that first. The agent only works on symbols you are actively monitoring. Once the watchlist is populated, the Thesis Drift Agent appears with ranked cards, score, reasons, linked signals, and the Thesis Memory panel on the right.

Step 1: Build A Watchlist You Can Actually Monitor

A good watchlist is not a wish list of everything you vaguely like. It is a working list of names where you are prepared to update your view if new evidence appears. If you overload the watchlist with too many casual ideas, the agent will still rank them, but the output will feel noisy because your own attention is noisy.

A strong starting pattern is to keep a mix of held positions, near-buy candidates, and one or two names you are actively re-underwriting. Use the add flow in the watchlist panel on `/live`, and remove names that no longer deserve active monitoring. The more intentional the list, the more valuable the ranking becomes.

Step 2: Save Thesis Memory Before The Market Forces You To

Click any ranked card and use the Thesis Memory panel. This is where you save the core thesis, the break condition, the watch items, the horizon, and your conviction level. If you skip this step, the agent falls back to baseline coverage, which is still useful, but less tailored to what you personally care about.

The right time to write a thesis is before the stock becomes stressful. Calm conditions produce cleaner thinking. If you wait until a drawdown, a management surprise, or a spike in valuation, you are more likely to write emotional notes instead of durable decision criteria.

Step 3: Write Inputs Like An Investor, Not Like Marketing Copy

The best core thesis is short, specific, and falsifiable. “Great company in a big market” is too weak to help the agent. “Deposit franchise stays strong, cost of funds remains controlled, and management avoids reckless growth” is much better because it creates something that can drift.

Break conditions should describe what would force a re-underwrite. Watch items should describe what you want to keep tracking even if the thesis is still intact. For example, a compounder might have watch items like margin resilience, pricing power, and capital allocation discipline. A cyclical or turnaround name might focus more on capacity utilization, deleveraging, order flow, or promoter behavior.

Step 4: Choose The Right Horizon And Conviction

The horizon selector changes how you mentally interpret the output. A swing idea, a quarterly review name, and a long-duration compounder should not all be judged with the same patience. Set the horizon to the role the stock plays in your process, not the role you hope it will play someday.

Conviction is equally important because it reminds you how much evidence you should require before changing your mind. A low-conviction watchlist idea can move quickly from curiosity to rejection. A high-conviction core holding deserves deeper investigation before you conclude the thesis is broken.

Step 5: Read The Ranking Like A Triage Board

Each card is designed to answer one question: what should I look at first? `Re-underwrite` means the agent sees enough tension between the saved thesis and current signals that you should revisit the position with intent. `Watch` means there is something worth monitoring, but not necessarily a broken case. `On track` means there is no urgent evidence of material drift right now.

The score, summary, matched signals, and top reasons should be read together. Do not overreact to the score alone. The number is useful because it creates ordering, but the reasons tell you whether the drift is about price damage, news flow, Street view, or a thesis-specific mismatch.

Step 6: Open The Latest Signal Before You Open Full Analysis

When a card surfaces `Recent linked signals`, click `Open latest signal` first. That gives you the freshest event context before you ask deeper questions. This is important because many bad investing decisions come from reacting to a score without reading the evidence that produced it.

The point of the first click is not to prove the agent right. It is to understand the exact event that is creating drift pressure. Sometimes that signal confirms a real issue. Sometimes it is temporary noise. The workflow is much better when you inspect the event before building a narrative around it.

Step 7: Use The Analysis Workspace For Second-Order Questions

After reading the latest signal, click `Open analysis` to move into [Arthalekh App](https://arthalekh.com/app). This is where the Thesis Drift Agent hands the stock to a deeper reasoning workflow. The ranking tells you where to focus; the analysis workspace helps you explore what changed, how much it matters, what still holds, and which uncertainties remain unresolved.

A useful pattern is to ask second-order questions instead of headline questions. Not just “why is the stock down,” but “which part of my thesis does this event threaten, what evidence would confirm that threat, and what would have to improve before I add rather than trim?” That is where the product becomes a real investing assistant rather than a dashboard.

How To Use Notifications Without Creating Noise

Inside the Thesis Memory panel you can enable agent notifications, set an email, and choose whether to be alerted for `Re-underwrite only` or `Watch and re-underwrite`. For most serious investors, starting with `Re-underwrite only` is the better default. It keeps the inbox focused on names where the attention cost is clearly justified.

Notification settings are tied to the session watchlist, and the product keeps recent notification history so you can see why an email was sent. That audit trail matters. Alerts are only useful when you can later inspect whether the threshold, summary, and top reason were actually decision-worthy.

A Good Daily Operating Rhythm

A clean daily routine is short. Open `/live`, scan the Thesis Drift Agent ranking, open the newest linked signal for one or two names, and stop if nothing important changed. The product is at its best when it reduces compulsive checking instead of encouraging it.

The heavier work should happen on scheduled reviews. Use a weekly review to clean up watchlists and refresh weak thesis notes. Use a quarterly review to rewrite thesis memory after earnings, guidance shifts, or management commentary. That rhythm keeps your saved context current without turning every market move into a research project.

Example: How To Use It For A Compounder

Suppose you own a long-duration compounder because you believe in pricing power, clean capital allocation, and durable free-cash-flow conversion. Your break condition might be sustained margin damage plus evidence that management is chasing lower-quality growth. Your watch items might include distribution expansion, working-capital discipline, and market-share quality.

If the agent moves that stock to `Watch`, you probably do not need to act immediately. If it moves to `Re-underwrite` and the top reasons mention margin stress, negative trend context, and a linked signal that questions capital discipline, that is the moment to open analysis and test whether the thesis is merely wobbling or actually weakening.

Example: How To Use It For A Fragile Position

Now imagine a cyclical or turnaround name that you do not yet fully trust. In that case, the agent is even more valuable because your thesis usually depends on a small number of conditions staying true. A disappointing update on demand, leverage, order inflow, or execution can change the risk-reward much faster than it would for a mature compounder.

For these names, a `Watch` status can already be meaningful. The right move is not always to sell instantly. It is to recognize that fragile theses deserve faster evidence review, tighter break conditions, and less storytelling tolerance than premium-quality holdings.

Common Mistakes That Make The Agent Less Useful

The first mistake is saving generic thesis notes that could apply to any company. The second is writing break conditions so vaguely that nothing could ever trigger them. The third is leaving old watchlist symbols in place long after they stopped mattering. The fourth is turning on low-threshold notifications before your thesis memory is strong enough to separate real drift from ordinary noise.

Another common mistake is using the ranking as a substitute for valuation work. The agent can tell you where the thesis needs attention. It cannot decide whether a stock is cheap enough, expensive enough, or sized correctly for your portfolio without your judgment.

What The Thesis Drift Agent Should Not Do For You

It should not place trades, remove the need for reading, or replace your valuation framework. It should not turn temporary volatility into fake urgency. And it should not become an excuse to monitor dozens of low-conviction names just because an AI system can technically score them.

Used well, the agent does something more practical. It protects attention. It keeps your thesis visible. It shortens the distance between a new signal and a structured review. That is exactly what most investors need in the agentic AI era.

Final Checklist Before You Rely On It Daily

Make sure the watchlist only contains active names. Make sure every important holding has a real thesis, a real break condition, and a small set of watch items. Start notifications at the stricter threshold. Treat `Re-underwrite` as a prompt to investigate, not a command to act. Then use `/app` to do the slower thinking that should always sit behind a serious portfolio decision.