Signals considered

The funding announcement and investor information can be cross-checked
A specific TGE, token launch, or unlock date appears
The unlock schedule may create concentrated sell pressure
Reviewable discussion exists around community activity and industry reputation

An analyst at a market-making firm is preparing for the weekly meeting. He needs to give the team one conclusion: do the two market-making candidates added this week deserve human due diligence?

The conclusion should be supported by evidence: official funding disclosures, investor confirmation, the published unlock schedule, on-chain or public records, and reviewable community discussion. In practice, claims about those subjects are scattered across Telegram groups. He remembers that he “saw it somewhere,” but cannot name the source.

This is a familiar market-making problem: a poor decision can put inventory and reputation at risk, yet early claims are buried in unorganized group conversations. He used to rely on memory. If he remembered a discussion, he could find it again; if he did not, all he could say was, “Let me check.”

Today is different. Instead of opening the group list, he opens the TOP Prospect workspace. Candidate messages are already filtered and ranked, while every item still retains its own source, time, original wording, and context. He reviews the records, checks the external claims against public materials, and writes two recommendations: one candidate may enter human due diligence; the other should wait for evidence about token economics.

He enters the meeting with recommendations whose source trail and remaining unknowns he can explain.

This article follows one person: an analyst at a market-making firm and what changed after he began using TOP. It shows what the product organizes, ranks, and preserves—and what it never does for him. Judgment, due diligence, and approval remain human work.

NOTICE: The team, group messages, and business conditions in this article are composite illustrations used to demonstrate the product workflow. They do not represent a real customer, conversation, contract, revenue result, or conversion outcome.

Before: Three Problems That Made Him Nervous Before Meetings

Problem one: early claims are in groups, but the evidence is elsewhere.

A market maker evaluates many things: whether a funding claim is corroborated, who the investors are, what the published unlock schedule says, and what remains unknown about community quality. Group posts can point him toward those questions, but they are claims, not proof. Before the meeting, he often starts with “I think I saw a funding announcement somewhere” or “someone in one group criticized the token economics,” then must find the original post and check official project materials, the investor’s website, public token documents, and other appropriate sources.

Problem two: a wrong judgment can cost money and reputation.

A market maker may put inventory and market reputation at risk when accepting a project. He cannot guess when deciding whether a candidate deserves due diligence. But refusing to guess is different from having organized messages to review. He wants to verify the claims, but he first needs to find the original posts and the public materials that can corroborate them.

Problem three: the discussion changes quickly.

Relevant discussion can appear before TGE and may change quickly. The analyst needs to notice a candidate early without turning speed into certainty. A timely candidate record gives him a place to start; it does not shorten or replace due diligence.

Together, these problems mean: he wants to make a serious judgment, but the claims are unorganized and the evidence still requires human checking.

After: Three Things the Product Does for Him

First: Separate Project-Discussion Groups From Peer Groups So the Information Sources Are Right

He reorganizes the monitoring sources. TOP Prospect only processes groups he actively connects and is authorized to access. Groups where projects gather—Web3 project discussions, funding communities, and unlock or token conversations—become information sources. Market-maker peer groups become market observation, used for industry context but excluded from the project-evaluation queue.

The result: project announcements, unlock discussions, and peer commentary no longer enter one undifferentiated stream. Each source keeps its role, and the analyst can decide which message deserves review first. A source category still does not make its claims true.

The corresponding product capability is monitoring-source management. You choose the groups; the system does not join groups for you. Information sources and market observation remain separate, and peer groups do not enter the evaluation queue.

Second: Each Candidate Message Becomes a Reviewable Record

Instead of assembling “what do we know about this project?” from memory, he starts with records like this:

Field	Content
Business category	Potential market-making project (evaluation)
Candidate messages	Funding claim / unlock discussion / community activity / industry commentary, each kept as a separate item with original text and source
Source	Multiple connected groups · sender IDs
Time	Original timestamp for each message (UTC+8)
AI score	A 0–100 ranking score and “High Priority / Important / General” label calculated from factors such as the base score, signal strength, importance, number of matched keywords, recency, and repeated mentions
Rationale	Each item matches the configured project terms; whether two items concern the same entity still requires human confirmation
Status	Pending Evaluation

These records solve his first problem. “I think I saw this somewhere” becomes “these messages mention the candidate, and each source and time is here.” He can follow the links to the original messages instead of reconstructing the trail from memory. The score only tells him which record to review first. He still confirms whether the messages concern the same entity and whether any claim is true.

The corresponding product capability is extraction rules + lead organization. You define what counts as a project signal worth attention in your business language, such as funding, unlocks, TGE timing, or community changes. Keywords and semantic rules retrieve and rank candidate messages. Each item keeps its own evidence trail; a person must confirm whether names refer to the same project and check every external claim against authoritative sources.

Third: Give Every Candidate a Status So Evaluation Progress No Longer Depends on Memory

He uses the product’s existing manual statuses—New Lead, Pending Follow-up, Followed Up, or Invalid—to show whether a candidate record has been reviewed. When he marks one invalid, the reason can be captured in the team’s review process—for example, “funding claim unsupported,” “unlock document unavailable,” or “name collision with another project.” Entering due diligence can remain Followed Up; it is not a conversion.

At month-end, the team can inspect its review reasons and decide whether the current rules retrieve too many unsupported claims. An operator may then manually add an exclusion, change a keyword, or adjust a source priority. The recorded outcome informs a human rule change; the product does not learn from it or rewrite the rules automatically.

The corresponding product capability is status flow. The team updates status manually. The product does not read direct messages and does not know whether a contract was signed; it records the status selected by the team. Who performs due diligence and how the committee decides remain human choices.

One Review Path: From Two Messages to Two Due-Diligence Questions

Follow the composite handling process from beginning to end.

On Monday, two messages appear in two different groups:

Group A: “Project A says XYZ invested. Token launch may be next quarter.” Group B: “Does anyone have the official unlock document for Project B? I saw a claim about a large release next month.”

System assessment: Both messages match “funding / token launch / unlock / market making.” Each appears as a separate candidate item with its own project string, source, time, original wording, and rationale. The matches justify reviewing each record; they do not prove that the investment happened or that the unlock claim is accurate. If another group later uses the same project name, the analyst still has to confirm whether it refers to the same entity.

He opens the records: The original text, time, and source from both groups are present, with links back to the messages. He then checks the project’s official materials, the named investor’s website or announcement, and the latest public token or unlock documentation. If those sources do not corroborate a claim, he records it as unknown rather than filling the gap.

Before the meeting, he writes two recommendations in the system:

Candidate one: mark Followed Up and recommend human due diligence because the investor’s public disclosure and project materials can be checked
Candidate two: mark Pending Follow-up and request the official token-economics and unlock documents before making a recommendation

In the meeting, he says:

“Candidate One may move into due diligence because the named investor has a public disclosure we can review. For Candidate Two, I only found a group claim about the unlock; I recommend waiting for the official document before assessing sell-pressure risk.”

Every statement is separated into claim, source, and remaining unknown. When his manager asks about the funding, he can distinguish the Telegram claim from the investor’s public disclosure and show both. If the sources conflict, the record says so.

If Candidate One enters formal due diligence, its record can remain Followed Up. If Candidate Two never supplies an authoritative unlock document, the analyst can keep it pending or mark it invalid, while the team records “unlock claim unsupported” in its review process. Neither state says that the product verified the project or that a commercial outcome occurred.

During review, he records one manual change to consider: messages that include a link to an official token document may deserve earlier review than unsupported claims about unlock percentages. The analyst still reads the document and judges its implications; the link is a retrieval clue, not proof that the project is suitable for market making.

Return to the Question: What Did the Product Actually Do for Him?

He did not change industries, teams, or standards. He only changed where he spent his time:

	Before	After
Finding information	Reconstructs “I think I saw this somewhere” from memory	Reviews organized records with sources and timestamps
Meetings	When he cannot answer, he says, “Let me check”	Every conclusion has provenance and survives questions
Catching early discussion	Notices a claim only when someone remembers it	Sees a ranked candidate while preserving the original source
Result	Starts from memory and must reconstruct the trail	Starts from a reviewable record, then performs the same human due diligence

But the product never does three things for him:

It does not decide whether the project is good. The AI score only ranks. It determines which project appears first; whether the project deserves market making can only be decided after his due diligence.
It does not contact the project. The product does not read direct messages or send messages automatically. Contact remains his work.
It does not approve the project. He changes the status manually. The product does not know what the market-making committee ultimately decides.

The product organizes and ranks. Judgment, due diligence, and approval remain human work. That organization changes him from someone who answers from memory into someone who answers from records. He no longer has to feel uncertain before the meeting.

How a Signal worth attention is found

See how Top Prospect finds and organizes Signals worth checking, keeps the original Telegram context, removes duplicates, and helps you decide what to review first. You decide whether to follow up and what to do next.

Open the methodology and core definitions

Before the Weekly Meeting, His Manager Asked, “Is This Project Worth Market Making?”