AI Is Changing BD and Capture: Here's What It Looks Like

The revolution in AI has already been televised. The question now is what comes next, and what it means for the teams trying to win government contracts.

That was the framing Bruce Feldman brought to a recent NDIA/ETI webinar on how artificial intelligence is transforming business development and capture. With thirty years of GovCon BD experience and twenty-two years in the Air Force, Bruce doesn't approach this from a theoretical angle. He approaches it as a practitioner who has spent weekends in proposal rooms, managed capture at the executive level, and watched AI reshape what's possible in real time.

Here's what he covered, and why it matters for your team.

The Revolution Has Already Happened, Now Comes the Discipline

Generative AI has been adopted faster than the cell phone, the iPad, the PC, and the Internet. Over 80% of GovCon companies have adopted it in some form in roughly three years, despite navigating security reviews, governance frameworks, and significant organizational skepticism.

The productivity gains are real. Conservative estimates put efficiency improvements at 30% or more for specific tasks. Most of that impact has landed in proposal writing first, because AI writes competently and fast. But that's only the beginning .

What's ahead is evolution: incremental improvements, better tooling, deeper integration, and organizations learning to use it with real discipline rather than curiosity.

The Core Capability That Changes Everything

Before getting into use cases, it's worth naming the underlying shift. Prior to generative AI, technology was largely pattern-matching and machine learning: deterministic systems that recognized what it had seen before. Generative AI does something different. It takes disparate content from diverse sources, finds what's relevant, combines and synthesizes it, and produces coherent, high-quality output in response.

For BD and capture, that matters enormously. Capture managers carry an extraordinary load. They're expected to synthesize competitor intelligence, customer history, budget documents, prior proposals, and internal capability data, then distill it all into a coherent, credible win strategy on deadline. AI doesn't replace that judgment. But it can handle the research, synthesis, and drafting work that consumes most of the hours, which means people get their time back.

What "BD Domain Aware" Actually Means

There's an important distinction between using a general-purpose AI platform and using one built specifically for business development. When you ask ChatGPT or Gemini a question about your capture strategy, you get a generic answer. These platforms don't know what makes your company different. They don't have your proprietary past performance data, your win themes, your customer relationships, or your competitive positioning.

A BD domain-aware platform is different. It already understands the landscape: what a transition plan is, what goes into a SEMP, how to structure a capture gate review, and what evaluation criteria typically look like. That context is embedded before you ever submit a prompt. Combined with your proprietary data, it can generate content that actually reflects your company's position, not a generic template.

This distinction matters especially as the FAR undergoes its most significant overhaul in years, with new preferences for commercial products and outcome-based service delivery. The teams that can rapidly adapt how they evaluate and respond to opportunities, without waiting for people to update their mental models, will have a real edge.

Where AI Is Already Changing BD and Capture

Opportunity Identification and Qualification

AI can now read a solicitation announcement, compare it against your company's capabilities, and give you a preliminary bid/no-bid assessment in minutes. That's not a replacement for human judgment. It's a triage tool that replaces the hours of manual reading, flagging, and synthesizing that currently sit at the front of every BD workflow.

For large companies with two or three years of capture runway, that triage might feel optional. For companies responding to a delivery order with two weeks' notice, it's the difference between a confident decision and a wasted bid.

Beyond triage, AI can draft your gate review deck from the same input. First-draft briefings that currently take hours can be ready in minutes. That doesn't mean you ship the AI's draft. It means your team is editing and refining instead of starting from a blank slide.

Customer Intelligence and Assessment

Understanding your government customer is foundational to capture. AI can now pull from an enormous range of sources: agency websites, conference presentations, LinkedIn profiles, journal articles, and Congressional budget justifications. It synthesizes a picture of what the customer cares about, how they've evaluated work in the past, and what their next solicitation might emphasize.

Capture managers do this today. AI can do the same research in minutes and hand the capture manager a synthesized brief to refine, challenge, and build on.

With the advent of AI agents this research can happen automatically. An agent can be triggered to collect, synthesize, and structure customer intelligence without waiting for a human to kick it off. The capture manager's job shifts from doing the research to working with its output.

Competitor Analysis

SEC 10-K filings, earnings calls, staffing patterns, contract awards: the data is out there. Most teams don't have the bandwidth to monitor it consistently across ten competitors. AI does. Setting up an agent to continuously collect and synthesize competitor data, flagging relevant changes for human review, is now well within reach.

Solution Development and Win Strategy

The highest-value application of AI in capture may be in the workshop setting. Before an RFP drops, you can ingest historical RFPs from the same customer, relevant agency priorities, and your own capability data, then ask the AI to forecast likely evaluation criteria, suggest differentiating strengths, and identify capability gaps that need to be addressed.

The humans in the room review what the AI generates, challenge it, refine it, and turn it into a real win strategy. The AI accelerates that process and surfaces things the team might miss.

What's Coming in the Near Term

AI Agents. Expect broader adoption of agents that automatically invoke research, synthesize data, and structure outputs without needing a human to prompt every step.

Multi-Model Systems. The next frontier is coordinated systems where multiple specialized models work in parallel: one for research, one for competitor analysis, one for drafting, coordinated by an orchestrating model. Cost efficiency is part of the appeal: you use the expensive, high-capability model where it matters most.

Knowledge Graphs. Knowledge graphs understand not just entities but the relationships between them. Who worked where before? Which contracting shop issued this RFP, and what's their history? That relational context significantly reduces hallucinations and improves response quality.

Predictive Analytics. Analytics-enabled platforms can start to provide more rigorous, data-backed probability assessments and help teams prioritize which bids are actually worth pursuing.

The Risks Worth Taking Seriously

None of this comes without caveats.

Hallucinations are real. When AI makes up citations, invents legal precedents, or misattributes capabilities, it's not a minor annoyance. It's a contractual and legal risk. Proposal documents are binding. Government source selection boards are scrutinizing AI-generated content. The solution is high-quality input, human review at every critical gate, and platforms purpose-built to minimize hallucination through domain-aware design.

Human oversight is not optional. AI works for your team, not the other way around. The most productive model is AI as an enabler, handling research, synthesis, and drafting, with human beings making the decisions that require strategic thinking, institutional knowledge, and judgment.

Make vs. buy decisions deserve scrutiny. Companies that attempt to build their own in-house BD AI platforms face significant hidden costs: organizational change management, ongoing maintenance, continuous model updates, and diversion of your top people into platform development rather than winning work. The data suggests roughly 80–90% of companies that attempt internal large language model builds abandon the effort within two years.

The Bottom Line

If you are competing against someone who is AI-proficient and you are not, you are at a disadvantage. That gap is already real and it's widening.

The good news is that the tools, platforms, and approaches are mature enough to produce genuine results today, not in some future state. Opportunity triage, customer research, gate review drafting, solution workshops, competitor intelligence: all of these are areas where AI is delivering efficiency gains for teams using it well right now.

The teams that will pull ahead aren't the ones experimenting with prompts. They're the ones building AI into how they actually work, with the right platforms, the right data, and human judgment firmly in the loop.

Watch the full NDIA/ETI webinar to see how Procurement Sciences supports the entire BD lifecycle.