# Auto, Leads & Prospecting Leads is where a company builds its own prospect lists. You describe the kind of company or person you want to reach in plain English, Neotask researches them, and the result is a structured, reviewable table saved to that company. Every list belongs to one company. Leads captured for one company are never visible to another, and never leave the tenant that captured them. This guide covers the first run, day-to-day searching, reading and exporting a list, how billing works, and how your autonomous agents can use the same system. --- ## Where Leads Lives Open a company, then select the **Leads** tab. What you see depends on whether that company has used Leads before: - a company that has never run a search opens the **first-run setup** - a company that has opens the **Leads workspace** Leads is per company. Turning it on for one company does nothing to the others. --- ## First Run: Setting Up Lead Finding The first time you open Leads for a company, Neotask offers to set up lead finding for it. This is a full-screen flow, and it exists so your very first search already knows who the company is and who it sells to. The setup moves through these stages: 1. **Preparing**: Neotask reads what the company already knows about itself: its profile, its plan, its goals, and anything captured during company onboarding. 2. **Review**: you see the prospect profile it drafted: what the company sells, who it sells to, and the kind of buyer worth targeting. Everything here is editable. Correct it now and every later search starts from the corrected version. 3. **Running**: your first search runs immediately using that profile. 4. **Complete**: the workspace opens with your first lead table in it. ### You Can Leave Once the first search starts, it runs on Neotask servers, not in the window. You can close the page, switch companies, or quit the app. The search keeps going and the results are waiting when you return. Because of that, there is no cancel button during the first search. The action offered is **Run in background**, which simply lets you leave. ### Correcting The Profile Later The profile is not frozen after setup. Say what changed in the chat and the next search uses the correction: > We stopped selling to agencies. Focus on in-house marketing teams at B2B SaaS companies. --- ## The Leads Workspace The workspace has three panels. ### 1. Conversation rail (left) Every lead conversation for this company, newest first. Select one to reopen it with its full history and the lists it produced. Starting a new conversation does **not** start the company over. The company keeps one lead workspace, so a new conversation reuses the same prospect profile and the same captured history. Ask for something different, not something disconnected. The rail collapses when you want the table full width. ### 2. Results pane (middle) Five views, described in detail below. ### 3. Chat (right) Where you describe what you want. The message box is always available, so you can queue a refinement while a search is still running. Above the input, suggestion chips offer useful next actions based on what the company already has. They float above the composer and never push the conversation around. --- ## The Five Views ### Overview Orientation for the whole company: - **Searches**: how many lead searches this company has run - **Lead tables**: how many captured result tables it holds - **Prospect rows**: total rows across every table - **Unique leads**: rows after duplicates are collapsed - **Recent searches**: the latest runs and their outcomes - **How lead finding is billed**: the current allowance and what a search draws from ### Searches Every search this company has run, with its brief and its state: | State | Meaning | |---|---| | **Running** | working now; results appear as they are captured | | **Completed** | finished and captured to a lead table | | **Succeeded** | finished and returned results | | **Failed** | stopped on an error; nothing captured | | **Cancelled** | stopped by you or an agent | Select a search to see its brief, its runs, and the table it produced. ### Prospects Every captured lead for the company in one place, across all lists. This is the view to use when you want to answer "do we already have this person?" before spending anything on a new search. ### Lead tables The captured result tables, newest first. One finished search produces one table. ### Runs The individual execution attempts behind each search, including continuations. Useful when a search was refined several times and you want to see what each pass added. --- ## Running A Search Describe what you want in plain English. Specific briefs produce better tables than broad ones. **Weak brief** > Find me some leads **Strong briefs** > Find heads of marketing at B2B SaaS companies in the US with 50 to 200 employees that recently raised a Series A. > Find operations managers at independent dental practices in Texas with more than three locations. > Find agency owners in the UK who publish case studies about paid social and have a team page listing at least five people. If a brief is too vague to run well, Neotask tightens it with you first rather than spending on a weak search. ### Refining Instead Of Restarting To adjust a search, continue it rather than starting a new one. Continuing keeps the rows already captured and builds on them. > Same list, but only companies with in-house engineering teams. > Add a verified work email for each of these rows. > Drop anyone without a LinkedIn profile, then find twenty more like the rest. Starting fresh when you meant to refine discards context and costs more. ### While A Search Runs - Results appear as they are captured; you do not wait for the whole run - You can leave the page, the company, or the app - You can queue the next instruction in chat - You can stop a running search from the Searches view --- ## Reading A Lead Table Lead tables use the same table as Tasks and Goals: the same sorting, the same column behavior, the same horizontal scrolling for wide tables. Scroll sideways to reach every column; nothing is hidden. **Click any row to open the full lead detail.** The table shows a summary. The detail panel shows everything captured for that lead, including fields too long for a cell. ### Columns You May See Research returns whatever it finds, so tables vary. Neotask recognizes common fields and renders them appropriately: | Group | Fields | |---|---| | Person | name, title, photo | | Company | company, website, about | | Contact | email | | Social | linkedin, twitter, facebook, crunchbase | | Assessment | fit, signal, evidence, confidence, why, status | Anything unrecognized still appears as a normal column. Nothing is dropped for not matching a known shape. ### Duplicates If a lead already exists in another list for this company, the row is flagged with an information icon showing where else it appears. **Duplicates are flagged, never removed.** A row stays in the list it was captured in, because a list is a record of what a specific search found. Removing rows would silently rewrite that record. When you want a clean file, use **Export CSV without duplicates**. --- ## Exporting Two options from a lead table: - **Export CSV**: downloads the table. Column order and column labels are exactly as stored, so the file matches what you reviewed. - **Export to Company Files**: saves the table into that company's Files, where agents, tasks, and other company work can use it. Use Company Files when the list should feed further automation. Use CSV when it is going to a person or another tool. --- ## Billing **The first 25 leads each month are included.** After that, delivered leads draw from your Neotask credit balance. The Overview panel always shows the current position: how many are included, how many are used, how many remain, and when the allowance resets. Trust that panel over any number written here, because the allowance is set on our servers and can change without an app update. Two details worth knowing: **The allowance is included leads, not a cap.** Running past it does not stop lead finding. It bills. **Lead finding always uses Neotask credits.** It runs on Neotask servers, so it uses Neotask credits in every AI mode, including when your assistant is running on your own sign-in or your own API key. Bringing your own model key changes where your *assistant* runs. It does not change where lead research runs. A search that returns nothing still represents work performed, so usage is metered on the work rather than counted per row. Cost per lead is therefore an average that moves with how hard a brief is to satisfy. Tight, well-specified briefs are cheaper per useful lead than broad ones. --- ## Using Leads With Agents Leads is fully available to autonomous agents and company orchestrators, per company and per item. An agent can check what the company already holds, run a search, refine it, and read the results, all inside the same governed system you use. ### Free reads These never spend. They only read this company's own records: | Tool | Purpose | |---|---| | `leads_overview` | orientation: counts and recent searches | | `leads_prospect_search` | search already-captured leads across every list | | `leads_lists` | list the captured lead tables | | `leads_list_read` | read one captured table | | `leads_search_status` | check searches, poll a running one | ### Spending actions These do real work and **require human approval every time**: | Tool | Purpose | |---|---| | `leads_search_start` | start a brand new search | | `leads_search_continue` | continue an existing search: refine or enrich it | `leads_search_cancel` stops a running search. Rows already captured are kept. The read tools and the spending tools are separate by design, so an agent can look before it spends and can never reach a spending action through a read. ### Expected agent behavior A well-behaved agent checks stored records before starting anything new, tightens a vague brief before running it, and continues an existing search instead of starting a second one for the same goal. The built-in lead-finding skill enforces exactly that sequence. ### Automating It Because the tools are available to the planner, lead finding can run as scheduled or triggered work: > Every Monday, check our lead lists and find twenty new prospects matching our best-fit profile, then add them to Company Files. The spend approval still applies. A scheduled search asks before it spends. --- ## What Leads Does Not Do Leads builds and enriches lists. It is deliberately not an outreach tool. - It does not send email, messages, or connection requests - It does not publish anything or write public copy - A captured list is an internal record for your team to review Export the list and use your own outreach tooling to contact anyone on it. --- ## Common Questions **Does a new conversation start the company over?** No. One company keeps one lead workspace, one prospect profile, and one captured history. A new conversation reuses all of it. **Can I close the app while a search runs?** Yes. Searches run on Neotask servers. Close the page or quit the app and the results are waiting when you return. **Why is a lead in two lists?** Because two searches both found it. It is flagged, not removed, so each list stays a true record of what that search found. Use **Export CSV without duplicates** for a clean file. **Why did a search return fewer rows than I asked for?** The brief was narrower than the available data. Broaden one constraint at a time, continuing the same search rather than starting over. **Does my own API key make lead finding free?** No. Lead finding runs on Neotask servers and uses Neotask credits in every AI mode. **Can I change the monthly included amount?** It is set on our servers, not in the app. The Overview panel always shows the current figure. --- ## Related - [Auto, Company Dashboard](auto-companies-dashboard.md): where the Leads tab lives - [Auto, Apps, Integrations & Files](auto-companies-apps-integrations-and-files.md): where exported lists land - [Auto, Approvals & Safety](auto-companies-approvals-and-safety.md): how spend approvals work - [Billing](billing.md): credits and balances - [Usage Rates](usage-rates.md): how usage is metered