Deal Sourcing · Defense Investment
AI that does the research and puts firm knowledge to work.
A custom software platform for deal sourcing and company intelligence, with an agentic layer on top. The AI agent captures incoming information and answers questions from firm data, while an AI research engine prepares company profiles—reducing the work between finding an opportunity and evaluating it.
- Organization
- Defense Tech Investment Firm
- Platform
- Custom Software + Agentic AI
- Scope
- Deal Sourcing, Company Intelligence
- Focus
- AI Capture, Research & Retrieval
The Outcome
Less work entering, researching, and finding information.
01 · Before
Information kept arriving. The full picture was hard to find.
Company information was scattered across business cards, brochures, emails, and notes. Relationship context often stayed with the person who held it. Research meant reconstructing the story, and departures could take valuable knowledge out of the firm.
02 · The Approach
Give AI the preparation. Keep people on the decisions.
Analysts were spending time entering details, collecting background research, and reconstructing context. The AI handles that preparation so the team can spend more of its time assessing companies and developing relationships.
Where Automation Adds Value
Extracting useful information from incoming material, profiling companies, and retrieving relevant context from the firm’s records.
Which companies to pursue, what to invest in, and how to develop each relationship remain decisions for the team.
The return comes from less manual entry, less research from scratch, and less searching for answers—not simply having another place to store records.
03 · The System
An AI agent that works with the firm’s intelligence.
The custom platform brings company records, relationship history, documents, and research into one system. Its agentic layer uses the platform’s tools to read uploads, retrieve intelligence, and prepare updates for approval. Automated research adds company profiles to that shared foundation.
- 01 / 04
Let the Agent Handle the Data Entry
Send a business-card photo, brochure, or notes to the agent. It extracts the details, searches for the company, and prepares record changes for confirmation. The team reviews the result instead of transcribing the source material.
From a business-card photo to a proposed update
Read Extract contact and company details Find Search for the relevant company Prepare Show proposed additions to the record Review and confirm before saving - 02 / 04
Start With AI Research, Not a Blank Profile
The AI research engine reviews company websites, registry data, and technology readiness, then writes a summary. Analysts can begin with a prepared profile and focus their attention on evaluating the opportunity.
A consistent foundation for evaluating a company
Website Company and technology overview Registry Company registration information Readiness Technology readiness research Summary A written company profile Research and source material available in one system - 03 / 04
Get Answers Without Rebuilding the Context
Ask about companies, technologies, contacts, or past discussions. The agent searches records, documents, notes, and interactions to answer from the firm’s own data. Captured knowledge stays usable when people leave, reducing the need to repeat research or rely on whoever remembers.
“What do we know about the counter-UAS companies in our pipeline?”
Companies Relevant profiles and technology research People Contacts and team relationships Context Documents, notes, and interaction history Answers drawn from the firm’s records - 04 / 04
Find the Relationships That Need Attention
Ask the agent who has not been contacted since an event. It searches interaction history kept current by email and calendar sync. Activity signals and shared watchlists help the team prioritize follow-up without checking each relationship by hand.
“Who haven’t we followed up with since the conference?”
Find Contacts and companies recorded at the event Check Emails and meetings since the first conversation Surface Relationships with no follow-up recorded
04 · Human Control
The agent does the preparation. Users approve the changes.
The assistant can only use tools permitted by the user’s role. Before logging an interaction or editing a record, it presents the proposed change and waits for that user to confirm it.
Agent Systems Handle
- Reading uploaded cards, brochures, and notes
- Researching and profiling every company
- Answering questions from the firm's own data
- Proposing record updates and follow-ups
People Own
- Investment decisions
- Approval of each proposed record change
- Which relationships to pursue
- Role permissions and access
The same capabilities are available to external AI assistants through MCP (Model Context Protocol), so the team can use the firm’s intelligence from the tools it already works in.