Agent471, the AI agent built into the Verity471 cyber intelligence platform, is now available. It tackles the research behind every request, with analyst tradecraft built in, so your Security team gets time back for the analysis and judgment that only they can provide.
Today we’re launching Agent471, the AI agent built into Verity471. It handles the legwork of intelligence production, from searching and compiling to citing and surfacing gaps, so analysts can focus on what the evidence means.
When we talk with Security teams, we hear the same challenge again and again. Requests are growing faster than the team. Leadership wants a threat landscape. The SOC needs a profile of a new actor. The fraud team has sent a request for information about a potential insider threat. All of these requests are compounded by adversaries using AI to accelerate intrusions. Defenders have less and less time to respond.
Producing intelligence is hard work. Analyst time is the bottleneck that we built Agent471 to solve.
Meet Agent471
When teams get a request from leadership, the SOC, or other stakeholders, the first things they need to determine are: “What do we know about this actor, and should we be worried?” Normally, that starts an extensive search. Analysts must sift through disparate intelligence, chase aliases, pull reports and posts together, and only then start working out what it all means.
Agent471 takes on that legwork. Your analyst asks the question in plain language, and Agent471 searches across Intelligence, Exposure and Hunt modes within Verity471, correlates what it finds and returns one sourced answer. There’s no query syntax, no switching between modes and no need to know the right alias first.
From there, analysts can check every step. Answers drawn from reports cite the title, date and exact passage used. Clicking a citation opens the underlying report, chat message or forum post in a side-by-side reader, without leaving the conversation. Each reasoning step and tool call appears live as Agent471 works, so its reasoning is never a black box.
When the actor profile or threat landscape report is ready, Agent471 can rewrite it for the SOC, the CISO or leadership. One round of research serves every stakeholder.
Analysts can also choose the depth each request needs:
- Instant model: summaries, rewrites, IOC extraction and lookups.
- Thinking model: actor research, IOC triage, vulnerability assessments and comparisons.
- Plan Mode: large, multi-part requirements. Agent471 scopes the question and drafts a collection plan the analyst approves before anything runs. More on how it works below.
Built-in tradecraft with Plan Mode
Some requests are bigger than others. A threat landscape comparison across several actors, or a brief that spans multiple intelligence domains isn’t one question, but many. An experienced analyst would start answering those by scoping the requirements and planning the collection. Plan Mode operates the same way.
- Scope. Agent471 asks scoping questions before it searches. The most important is usually the time window.
- Plan. It drafts an intelligence collection plan in the side-by-side window. The question is broken into narrower tasks that identify the relevant Verity471 sources answers depend on. The plan concludes with the gaps Agent471 expects before the search is run.
- Edit first. The analyst can review and edit the collection plan before the search runs: tightening the window, dropping a task or adding context about the organization.
- Subagents run the search. Tasks are handed to subagents that work through them in parallel where they can.
- Report. Findings return to the original thread with inline citations. The output transparently states where a task can’t be fully supported.
Analysts refine the collection plan before the search. The side-by-side window shows the requirement, collection tasks and sources, and identifies expected collection gaps.
The result is a plan your team shaped with transparent research steps it can stand behind. Beta teams found the scoping step made the output sharper, not just faster. The logistics company’s analysts said the clarifying questions pushed them to set tighter collection windows, such as the last 14 or 30 days, before any research ran.
The refinement layer: Guardrails in the background
What makes Agent471’s answers trustworthy is what happens before and after the AI model reasons over anything.
Agent471 starts with Intel 471 adversary intelligence, collected from inside the communities where threat actors collaborate and plan, not just from the open web. Before the agent sees that intelligence, Intel 471’s refinement layer checks each user’s access, screens the retrieved reporting for manipulation, summarizes it and attaches its source. The source behind each finding comes from Intel 471’s own records, not from the model’s memory, so the model never has to reconstruct where something came from.
After the agent answers, a citation guard checks every link against what Agent471 actually retrieved and removes anything it didn’t. When Agent471 can’t find something, it says so instead of guessing, and every answer is labeled as AI-generated, with sources to verify.


Stay in the chat with a side-by-side view of sources for verification, context and further investigation.
Agent471 is built for enterprise trust and Intel 471 does not use customer data to train models. Agent471 is read-only, checks each user’s entitlements on every request and encrypts data at rest. Conversations are visible only to the person who started them.
Agent471 isn’t autonomous; instead, it amplifies your analysts and gives them back time to focus on judgement calls that matter to the organization. Your team sets the direction, approves the plan and makes the call on what the findings mean.
How customers are using Agent471
No two requests look alike. One week it’s a quarterly threat landscape for leadership. The next, it’s a one-line question from the fraud team that needs an answer by the end of the day. What they share is the legwork, and that’s where Agent471 delivers. In beta, the biggest gains came from the work that often takes the longest.
Strategic threat landscapes: Evaluating threats to a sector, region or subsidiary can tie up an analyst for a day or more. It’s a natural fit for Plan Mode where the analyst shapes the collection plan, Agent471 does the collection, and the findings come back together, closing with a summary table.
Executive and stakeholder briefings: Weekly CISO briefings and daily intelligence reports never stop coming. Agent471 drafts them from cited findings in scope and rewrites them for each audience. One Fortune 50 retailer’s team found they could paste its output straight into their daily and weekly reports with minimal editing. For lean teams supporting large organizations, that matters.
Requests for information: Ad hoc questions and investigations from internal security teams rarely arrive on schedule. Agent471 gives analysts a faster path from a one-line question to a sourced answer they can send back.
Threat actor research: Analysts used Agent471 to build actor profiles, track personas and get up to speed on emerging or obscure groups, pulling actor tactics, techniques and procedures (TTPs), malware and victimology into one answer.
Underground chatter around critical vulnerabilities: When a new zero-day or n-day vulnerability is disclosed, the first question is whether threat actors have weaponized it. For one customer, Agent471 compiled 14 actor personas and their exploit claims across Telegram channels in about two minutes. The team then clicked through to the original posts to validate the findings before reporting them.
Custom risk scoring: Teams scored threats against their own internal criteria and were able to build country-level risk assessments using Intel 471’s geopolitical intelligence to help prioritize what mattered most to their organization.
Try Agent471
Security teams don’t need another chatbot. They need help with the work that stands between a request and an answer. Agent471 brings three things together to deliver it:
- Time back for your analysts: One question searches across Intelligence, Exposure and Hunt and returns a cited answer that’s ready to share with every stakeholder.
- Tradecraft built in: Plan Mode scopes the request and builds a collection plan your team approves before any research runs.
- Intelligence you can trust: Every answer draws on Intel 471 adversary intelligence that has been access-checked, screened and sourced, and Agent471 tells you when it can’t find something rather than guessing.
Agent471 is available now. Ask your account team to turn it on, or book a 30-minute walkthrough to see it in action.

