Posts by Chelsea Roberts
The Pilot Trap: Why an SDVOSB Founder Refuses the VA’s Favorite Word | Michael Schmitt
He spent half a million dollars of his own money, put his platform inside the VA’s own cloud, and then told the government he didn’t want a pilot. In this Success After Service episode, a 100% service-disabled Marine explains why refusing the pilot is the whole strategy.
In this Success After Service special episode of GovCon Unscripted, Mike Schmitt retired from the Marine Corps in 2004, spent years in medical devices at Medtronic, and then did something few small-business owners attempt: he self-funded a clinical toxicology lab, built his own laboratory information system, and stood it up inside the VA Enterprise Cloud through Syncly Direct so it could do imaging, lab, and AI in one place.
As a 100% service-disabled veteran running an SDVOSB, he learned the federal market the hard way. In this conversation with host Chelsea Roberts, he lays out the argument at the center of his approach: the pilot-and-funding cycle rarely produces anything, so instead of asking the VA for money and two years, he arrives already built, already authorized, and sells the result. He talks through the siloed VA and its VISNs, the records gap that leaves veterans carrying binders between systems, the credibility discount a small business faces against Deloitte and Oracle, the commercial dual-use of the platform in substance-abuse testing, and why, for him, this is about veterans taking care of veterans.
Timestamps:
00:00 Welcome and who Mike Schmitt is
01:00 From the Marine Corps and Medtronic into toxicology
02:00 Going in as a 100% service-disabled veteran (2018–2019)
03:00 Crossing the Rubicon: the MOU and interservice agreement
05:00 Building his own lab system and moving to AWS GovCloud
06:00 Syncly Direct inside the VA Enterprise Cloud
07:00 The records gap: a pregnant veteran and the PACT Act
09:00 The siloed VA and the VISNs
11:00 An agnostic platform: “Uber healthcare”
12:00 Lobbyists, the Hill, and why he walked away
13:00 The credibility discount: “if I’m not a Deloitte or an Oracle”
16:00 HIPAA and the compliance gap
20:00 Why he refuses the pilot
21:00 A nine out of ten, and the 2026 FORUM Innovation Award
23:00 Commercial dual-use: substance-abuse testing, the 97-compound panel
25:00 The heat map: reading opioid and health risk by region
29:00 Advice for transitioning veterans
31:00 Veterans taking care of veterans
33:00 How to reach Mike
Resources:
Other resources:
🎧 Listen now on:
🔹 Apple: https://podcasts.apple.com/us/podcast/govcon-unscripted/id1766384482
🔹 Spotify: https://open.spotify.com/show/1NL0kFAvmgECLTJpUZYNHy
Like and follow us:
🔹 YouTube: https://www.youtube.com/@GovConUnscriptedUS
🔹 Instagram: https://www.instagram.com/govconunscriptedpod/
🔹 Facebook: https://www.facebook.com/profile.php?id=61593788076719
🔹 LinkedIn: https://www.linkedin.com/showcase/govcon-unscripted/posts/
🔹 JOIN OUR #govcon #community: https://www.linkedin.com/groups/14618490/
It’s A Small World After All
Everyone in my feed is talking about it: the proposed changes to the SBA size standards.
On August 20, SBA published a proposed rule that rewrites small business size standards across the entire economy (91 FR 53741, RIN 3245-AI67). Comments close September 21. This is the third five-year review required under the Small Business Jobs Act, and SBA did not tinker at the margins. The agency collapsed 978 industry standards and 18 subindustry exceptions into 338 standards set at the 4- and 5-digit NAICS level. They shifted 208 industry groups onto employee-based measures and deleted the maximum size standard entirely (the old ceilings were $47 million in receipts and 1,500 employees). And they added a productivity adjustment on top of inflation for the first time in the program’s history. It also declined to lower any of the standards, including in the 45 industries where SBA’s own analysis said the number should come down.
The proposal moves the small business population from 6,344,967 firms to 6,459,508, a net increase of 114,541. For an administration that claims to want more small businesses, some might see that as a win. But almost nobody is sharing this number 37,002. That’s the number of firms that held contracts in FY2025, that would become newly eligible small businesses. Those firms accounted for 105,655 contracts worth more than $71 billion.
Those are not startups. Those are not new entrants. That is not even new competition, it’s seasoned contractors that get reclassified in a set-aside status.
Looking at where a lot of those companies land is helpful (to me anyway). There are 5,314 Engineering Services firms (541330) with current contracts that become small. 2,247 in Other Computer Related Services (541519). 2,171 firms in Custom Computer Programming (541511), 1,818 firms in Administrative Management and General Management Consulting (541611), and 1,663 firms in Computer Systems Design (541512) are also becoming small. These companies have active CPARS ratings, DCAA-compliant accounting systems, cleared staff, standing GWAC seats, and proposal shops that don’t have to ask what a Section L is. Many of them spent the last decade graduating out of small business status and building the infrastructure to survive without it. Now they walk back through the door carrying it all. Talk about punching above your weight!
SBA’s rule states that growing small businesses closest to the current thresholds will face the greatest competition from newly eligible firms. Those are the companies chasing the same work. If your firm sits at eighty percent of your current standard and your pipeline runs on set-asides, SBA just told you that you to plan on absorbing the impact.
But don’t worry! SBA also offered some reassurance. In one sentence across all 44 pages, SBA says the number of small businesses receiving federal contracts will not be reduced. In other words, the number of small business set-asides isn’t expected to be reduced. The competition for them is just going to go straight up!
Federal agencies obligated more than $883 billion in FY2025 and awarded a record $179 billion in small business prime dollars in FY2024. Unfortunately, widening eligibility does not create new requirements; there are no new appropriated funding dollars, and probably no additional awards. There won’t be additional funds to go to small businesses, just more “small businesses” for existing funds to go to. The requirement counts stay flat while the qualified bidder pool grows by roughly two percent nationally and far more than that inside the professional services codes where most of us live.
But fear not (kind of): The SBA highlighted the upside for contracting officers. The agency calculated that a shorter, simplified rule saves five hours and fifty minutes of reading time. That’s worth a whopping $219.57 per reader at June 2026 average hourly earnings, across roughly 37,600 contracting officers. That level of precision on reading time, in a rule that reshapes the entire small business industrial base, tells you something about which impacts got modeled and which ones got effectively ignored.
What about the defense?
For years, the size protest functioned as a competitive instrument. So did NAICS assignment arguments, and so did the exceptions that were carved out under specific six-digit codes. All three shrink under this proposal. Consolidating to 338 standards eliminates most of the gray area between adjacent codes (SBA uses shipbuilding versus boat building as its own example), and the rule removes every subindustry exception, including the information technology value-added reseller exception that many resellers have built their entire capture posture around. Size protests have already dropped from 500-600 per year in the 2011 to 2016 window down to roughly 300 annually between 2020 and 2024. Fewer standards and no exceptions mean fewer arguments left to make after award.
The conversions cut in both directions, too. If your standard moves from receipts to employees, a lean, high-revenue firm gains room. A staffing-heavy, field-services, or seasonally scaled firm can gain revenue headroom and still trip an employee count it never had to track before. Companies in these situations need to do some modeling of what these impacts would be, now, not during the holiday hiring surge.
This is why, if this rule passes, capture will be even more critical than it has been in the past.
Set-aside eligibility has never been a win strategy. It’s just a filter that determined whose proposal got evaluated. When that filter widens by 114,541 firms and 37,000 of them arrive with real past performance, that filter stops doing any meaningful work for you. What separates winners from participants is everything that happens before the solicitation posts.
Here is what I tell every client. Capture is the discipline of being known, credible, and specifically relevant to a named program office before that office writes the requirement. It’s reading the budget justification and the posture hearing testimony to find out what that customer is actually being held accountable for, then building your value statement against those mission achievement factors instead of your capability list. It’s knowing the incumbent’s weakest CPARS narrative. And it’s having your teaming decisions made before the sources sought hits, not scrambled together in the four days after.
Here are some things you can do in the next 90 days to posture:
- Model your size status under both the current and proposed standards.Pull your consolidated four-digit code, run your trailing five-year receipts and your trailing 24-month average headcount, and find out whether you gain headroom, gain competitors, or both.
- Rebuild your competitive set from SAM.gov and USASpending, not from memory. Filter FY2025 awards in your primary NAICS for firms that were other-than-small and will not be. That list is your new bid board reality.
- Move your effort left of the RFP. Reallocate proposal hours into pre-solicitation engagement, RFI responses that shape language, and industry day conversations where you learn the evaluation logic before it is written down.
- File a comment before September 21.SBA is asking, the docket is open, and the agency responds to specific operational details far better than to general objection. Silence here is a choice.
The set-aside is not going away. It is getting crowded, and crowded markets reward the firms that showed up early and knew the customer by name. If your growth plan for FY2027 depends on being eligible, you do not have a growth plan. You have a registration.
Never Just a Business Leader | Dr. Sarbari Gupta
She could have become a full-time executive years ago. She refused to stop being an engineer.
Twenty-five years later, that choice still shapes the company.
Dr. Sarbari Gupta founded Electrosoft in 2001, walking away from a steady job with nothing else lined up. Twenty-five years later the company runs around 200 employees, and she is still a technologist at heart. In this conversation with host Chelsea Roberts, she covers the leap that started it all, winning early contracts on relationships rather than pedigree, learning GovCon accounting from scratch, the difference between RMF/ATO and Zero Trust, pursuing CMMC Level 2 as a cybersecurity firm, learning to delegate, and the two things she has never been willing to hand off: the customer relationships, and her own connection to the technology.
Timestamps:
00:00 Intro
00:01 From electrical engineering to cybersecurity
01:00 The leap of faith: founding Electrosoft in 2001
02:30 Early challenges: GovCon accounting and learning to sell
03:30 Why federal — PKI, NIST and NSA connections
05:00 Has the compliance load grown? CMMC and FISMA
08:00 RMF/ATO vs. Zero Trust, explained
11:00 Barriers, focus, and growth
12:30 Scaling and the mid-2025 rebound
14:30 Learning to delegate at ~200 employees
16:00 The two things she will never hand off
18:00 Staying a technologist, and betting on AI
20:30 What is next for Electrosoft
22:00 Where to find Dr. Gupta and Electrosoft
Resources:
Other resources:
• Resources page on Electrosoft’s website
🎧 Listen now on:
🔹 Apple: https://podcasts.apple.com/us/podcast/govcon-unscripted/id1766384482
🔹 Spotify: https://open.spotify.com/show/1NL0kFAvmgECLTJpUZYNHy
Like and follow us:
🔹 YouTube: https://www.youtube.com/@GovConUnscriptedUS
🔹 Instagram: https://www.instagram.com/govconunscriptedpod/
🔹 LinkedIn: https://www.linkedin.com/showcase/govcon-unscripted/posts/
🔹 JOIN OUR #govcon #community: https://www.linkedin.com/groups/14618490/
Stop Running Your Company from Your Checking Account | Bud McDonald
Your bank account cannot tell you what next month looks like. A forecast can. Bud McDonald has spent 35 years proving the difference is the whole ballgame.
Most small business owners manage by their bank balance. Bud McDonald, founder of McMay Business Services, spent more than three decades leading turnarounds, managing P&Ls, and running large teams before he started helping small businesses do the one thing that separates the ones that grow from the ones that stall: see their money before it arrives.
In this episode, Bud and Chelsea get into the 13-week cash flow forecast, why a four-page business plan beats no plan, the receivables conversation every founder dreads, how to price the outcome instead of the hour, and the real signal that tells you when it is time to make your first hire. Practical, unglamorous, and exactly the kind of thing nobody hands you when you start a company.
Timestamps:
00:00 Intro
01:00 35 years of turnarounds, and why Bud started McMay
02:00 The business plan that fits on four pages
03:00 Budget, forecast, and the 13-week cash flow
04:00 The bank-balance trap
06:00 Chelsea’s spreadsheet lesson
07:00 Bookkeeping as the foundation
08:00 QuickBooks, Wave, and picking a scalable tool
10:00 The mistake behind most turnarounds: processes nobody uses
13:00 A turnaround story: communication first
20:00 Receivables, and the conversation nobody wants
22:00 When to fire a customer
25:00 Pricing: value your time, then price the outcome
30:00 There is no magic number for your first hire
34:00 Working on the business vs in the business
36:00 Where to find Bud
Resources:
Other resources:
🎧 Listen now on:
🔹 Apple: https://podcasts.apple.com/us/podcast/govcon-unscripted/id1766384482
🔹 Spotify: https://open.spotify.com/show/1NL0kFAvmgECLTJpUZYNHy
Like and follow us:
🔹 YouTube: https://www.youtube.com/@GovConUnscriptedUS
🔹 Instagram: https://www.instagram.com/govconunscriptedpod/
🔹 LinkedIn: https://www.linkedin.com/showcase/govcon-unscripted/posts/
🔹 JOIN OUR #govcon #community: https://www.linkedin.com/groups/14618490/
Is SLED Contracting Worth It for Federal Contractors? by David Fine
Over the course of a rocky FY26, every federal contractor has been hearing the same thing: diversify. Look at state and local. Stop betting the company on one buyer.
It’s good advice. It’s also incomplete, because “diversify into SLED” is not one decision. It’s a different decision for a highway contractor than for a software company, and the gap between those two cases is enormous.
So instead of arguing about it, we went and counted. What follows comes from open, live solicitations across state, local, and education buyers, the ones sitting on procurement portals and city websites this week.
How many state and local contracts are there in my NAICS code?
Start with the simplest question a contractor can ask. For every open federal opportunity in my code, how many open state and local ones are there?
Same advice, opposite outcomes, and the only thing separating them is which code you happen to work in.
Are state and local contracts smaller than federal contracts?
Yes, and by more than most people expect.
We compared the median published contract value in the same NAICS code on both sides of the fence:
A federal IT services firm accustomed to pursuing four-million-dollar computer systems design(opens in new tab) contracts will find the state and local equivalent sitting around a quarter of a million. That is a sixteen-fold difference in deal size, inside the same code, for what is often very similar work.
The gap narrows as the work gets more commoditized. Janitorial services(opens in new tab) runs less than two to one. Software publishers, under three to one. But the direction is consistent. The same capability sells for less to a city than to a federal agency.
That sounds like bad news. It isn’t, necessarily. It just means you have to be honest about what you’re buying. You are not buying a replacement revenue stream at your current deal size. You’re buying more shots, at lower stakes, with lighter competition, in a market where a small win still counts as past performance.
For a company trying to build a record rather than defend one, that trade is often worth making. For a company that needs to replace a $4M contract that just went away, it isn’t. Know which one you are before you commit a business development quarter to it.
How long do you get to respond to a state or local RFP?
Federal contractors bracing for a frantic scramble tend to be surprised here. State and local solicitations stay open a median of 29 days from posting to due date. Federal, 19.
You get about ten more days. The pressure in SLED is not the response clock. It’s everything that happens before the clock starts, which is the part nobody warns you about.
Who actually buys in the SLED market?
Ask a federal contractor to picture a SLED buyer and most describe a state agency. That’s not what the market looks like.
State governments account for under a quarter of open solicitations. Cities alone outnumber them. Add counties and K-12 school districts and you have close to sixty percent of the market sitting below the state level entirely.
This matters more than it sounds. If you’ve looked at state-level procurement and decided you know how SLED works, most of the market still looks nothing like what you saw. A city of forty thousand people might post a bid on its own website, give you a phone number, and award it in six weeks. Plenty of these buyers run a handful of solicitations a year, and the process reflects that.
How local does “local preference” actually get?
Every federal contractor eyeing SLED has heard that local preference exists. Almost nobody knows where the line gets drawn.
When a solicitation carries a local preference, it’s drawn at the city limit or the county line about two-thirds of the time. State and regional boundaries are the minority.
And the tighter the boundary, the harder the rule. Preference drawn at the city limit is roughly twice as likely to be a hard requirement, an eligibility gate rather than a scoring nudge, as preference drawn at the state line.
The practical translation: “we’re going after Texas” is not a SLED strategy. Being a known, registered, physically present vendor in three counties is a SLED strategy. The geography that matters is far smaller than most federal contractors assume, which is bad news for a spray-and-pray approach and very good news for anyone willing to pick a patch and own it.
How should a federal contractor start bidding on state and local work?
Four things, in order.
Look up your own ratio first. Before you commit any BD time, count open SLED solicitations against open federal ones in your primary code. If that number is under three, the diversification case has to be made on relationship risk, not volume.
Reset your deal-size expectations before your first pursuit, not after. Going in expecting federal contract values guarantees you’ll conclude SLED “isn’t worth it” after two bids.
Pick counties, not states. Then register as a vendor in each one, because most agencies search their own vendor database before anything gets posted.
Fix your discovery problem before your proposal problem. Your win rate is not what’s limiting you here. There is no SAM.gov for state and local work. The opportunities are scattered across tens of thousands of individual government websites, and the ones you never see cost you more than the ones you lose. You can start by pulling up the open solicitations in your own NAICS code(opens in new tab) and counting how many of those buyers you’d never heard of.
The federal contractors doing well in SLED right now aren’t better at government contracting than you are. They just did the arithmetic on their own code before they committed.
David Fine is the founder of Bidscope(opens in new tab), which tracks state, local, and education solicitations across more than 50,000 government sources and matches them to contractors by capability rather than by keyword or NAICS code. Bidscope also surfaces signals on upcoming opportunities, drawn from forecasts, published budgets, and expiring contracts.
The Capital Stacking Playbook: How Defense Founders Fund Hard Tech Without Losing Equity | Mollie Jahner
Hard tech is hard, and the money you raise to build it can quietly decide which customers you are allowed to serve. This week on GovCon Unscripted, Mollie Jahner makes the case that most defense founders reach for venture capital by reflex, when a stacked strategy of non-dilutive federal, state, and local funding, plus smart teaming, gets them further with more control.
Host Chelsea Roberts sits down with Mollie Jahner (Mission Cultivate, founder of Raise, formerly founder/CEO of Spring & Forge) on capital stacking, why the wrong investor can trip a CFIUS or foreign-ownership review and block your end customer, how teaming unlocks solicitations a single company cannot reach alone, and where Raise fits in helping founders find non-dilutive money faster.
Timestamps:
0:00 Intro
1:00 Why working moms make great operators
3:00 Getting government and industry in the same room
6:00 Founding Mission Cultivate and the Raise platform
10:00 Teaming and consortia: unlocking bigger contracts
13:00 The capital stack: non-dilutive funding beyond VC
15:00 The VC trap and the wrong kind of capital
18:00 Why human relationships beat AI noise
21:00 Who Mission Cultivate serves and the Raise launch
26:00 Wrap-up
Resources:
Other resources:
🎧 Listen now on:
🔹 Apple: https://podcasts.apple.com/us/podcast/govcon-unscripted/id1766384482
🔹 Spotify: https://open.spotify.com/show/1NL0kFAvmgECLTJpUZYNHy
Like and follow us:
🔹 YouTube: https://www.youtube.com/@GovConUnscriptedUS
🔹 Instagram: https://www.instagram.com/govconunscriptedpod/
🔹 LinkedIn: https://www.linkedin.com/showcase/govcon-unscripted/posts/
🔹 JOIN OUR #govcon #community: https://www.linkedin.com/groups/14618490/
The Money Isn’t Where You Think: How to Read Federal Contract Data | Eileen Kent
Everyone reaches for the free federal databases. Almost nobody reads them correctly, and the wrong read sends a whole pursuit in the wrong direction.
In this week’s episode of GovCon Unscripted, Chelsea Roberts sat with Eileen Kent, the Federal Sales Guide and President of Custom Keynotes, LLC, who has pulled the SAM ad hoc report thousands of times over twenty years.
She walks through the free public data that most contractors misread: why “place of performance” often shows the vendor’s address or the manufacturer’s plant instead of where the work lands, why “dollars obligated” is the only money figure worth trusting, why a string of negatives is usually an adjustment rather than a vendor being pushed out, and how to hunt for patterns of who buys what you sell from whom.
She makes the case that AI and paid aggregators can accelerate a bad decision because they cannot interpret dirty data, and that the deal is still closed in the field, through relationships. She closes on the discipline she cares about most: a real bid/no-bid process, and the strength to walk away when there is no relationship.
Timestamps:
00:00 Intro · 01:00 The Three-Step Program
02:30 Early tools (FPDS, USASpending, FedBizOpps, CCR)
06:30 Free vs paid · 07:30 The place-of-performance problem
12:00 Inside the SAM ad hoc report
15:00 Reading patterns (DHA, VA VISNs, CMOPs)
20:00 The 2025 shift of commodities to GSA
23:00 Why outreach bots fail
28:30 Dollars obligated and negative adjustments
33:00 Bid/no-bid discipline.
Resources:
Other Resource:
🎧 Listen now on:
🔹 Apple: https://podcasts.apple.com/us/podcast/govcon-unscripted/id1766384482
🔹 Spotify: https://open.spotify.com/show/1NL0kFAvmgECLTJpUZYNHy
Like and follow us:
🔹 YouTube: https://www.youtube.com/@GovConUnscriptedUS
🔹 Facebook: https://www.facebook.com/govconunscriptedpodcast/
🔹 Instagram: https://www.instagram.com/govconunscriptedpod/
🔹 LinkedIn: https://www.linkedin.com/showcase/govcon-unscripted/posts/
🔹 JOIN OUR #govcon #community: https://www.linkedin.com/groups/14618490/
The $50M Company That Almost Couldn’t Be Sold | Robert E. Jones
A $50 million company with 600 employees almost fell out of its own sale. Not because of revenue, but because every decision still ran through two people.
Most owners think exit planning starts when they call an M&A advisor. In this episode of GovCon Unscripted, host Chelsea Roberts talks with Robert E. Jones, CPA and Principal Advisor at Corvata, who has spent 20+ years in government contracting, and says that by then the value of the business is already fixed. The real work happens two to five years earlier, left of the transaction, where structure, discipline, and decision flow still can be changed.
In this conversation, Robert breaks down his Enterprise Readiness Operating Model, the seven decision failure modes that stall companies, why owner-dependency scares off buyers, and how tightening operations can move an EBITDA multiple from the low end of a 6 to 9x range to the high end. He also shows how tools most contractors already pay for can cut a 12-hour payroll run to two or three hours.
Timestamps:
00:00 Does anyone actually plan the exit?
01:00 “You will exit your business”: on your terms or not
02:00 Left of the transaction: where value is still changeable
03:00 The Enterprise Readiness Operating Model (structure, discipline, decision flow)
06:00 The seven decision failure modes
11:00 Due diligence: documents that “magically appear”
13:00 The $50M company that almost couldn’t be sold
20:00 Policy vs procedure vs work instructions
21:00 Getting off paper: SharePoint, Planner, and the tools you already own
25:00 Value growth: profitability, the multiplier, and the multiple
29:00 The 10-year math: working backward from the day you want out
33:00 The three kinds of optionality this work buys you
Resources:
Follow Robert E. Jones
Follow Chelsea Roberts
Other resources:
$30 Million, One-to-One Match, Five Agencies: Inside the Strategic Breakthrough Award
More thoughts on SBIR Strategic Breakthrough Awards
I keep getting lots of questions and requests about how to get one of these $30 million Strategic Breakthrough (SB) awards.
Well first, let’s break it down a few very pertinent facts:
SB awards are ONLY authorized (by PL 119-83) at the following SBIR agencies: DOD (DOW), HHS, NSF, DOE and NASA. None of the other SBIR agencies can even consider SB awards.
SB awards are FOLLOW-ON Phase II awards (based on the previously implemented Air Force STRATFI program). This means to be eligible, a company must already have an SBIR/STTR Phase II award.
Before finalizing an SB award, an eligible company must show evidence of a 1:1 match of the proposed award amount.
An SB award will come from the awarding agency SBIR budget. That is, PL 119-83 did not appropriate any new funds for SB awards, it simply allows SBIR dollars to be spent via SB without requesting permission.
MY THOUGHTS… (not facts!)
NIH & NSF have said they would likely use the new SB authority to expand their existing Phase IIB awards on a case by case basis. But both expected that IF they used the new SB authority, the award sizes would be more like < $5 million, not close to the $30 million maximum.
My personal view is that only the largest DOD components (AF, Navy, Army, MDA) will entertain anything near the $30 million maximum. Any SB award nearing $30 million would simply chew up too much of the overall SBIR budgets. Keep in mind that one $30 million award would be more than 10% of the individual NSF, NASA & DOE annual SBIR budgets.
So what should you do?
The four requirements will likely filter out most of those interested in pursuing an SB award.
If your firm DOES meet all the requirements above, then it is in your best interest to talk with your agency program staff to inquire how to pursue these larger awards.
Final thought:
This new authority gives the 5 agencies listed additional flexibility on how to use SBIR budgets to further their agency mission. I LIKE this new authority!! However, actual implementation of the authority will likely be limited in both scope (which agencies actually use the SB authority) and scale (the actual dollar amount of any resulting SB awards).
Connect with Jerry Hollister
BBCetc Website: https://bbcetc.com/
Additional Resources:
Public Law 119-83 DSIP (Defense SBIR/STTR Innovation Portal): https://www.congress.gov/119/plaws/publ83/PLAW-119publ83.pdf
SBA FAST Program: https://www.sbir.gov/community/fast
Design-Time vs. Runtime AI in Government: Why Most Agencies Get It Backwards | Doug Averill
Your AI pilot probably isn’t failing because the model is weak. It’s failing because you pointed it at the wrong end of the process.
Doug Averill has spent half his career in government and half in technology, and now serves as Vice President, Global Industry Market Leader, Government at Pega. In this episode of GovCon Unscripted, he draws a line most modernization teams never think to draw: the difference between runtime AI (burning tokens to execute the process you already run) and design-time AI (reimagining the process from the source policy before you build anything).
Apply AI at runtime to a broken workflow, Doug argues, and you pave the cow path: you carry the skipped steps, the noncompliance, and the suboptimal outcomes straight into the new system, only faster. He and Chelsea get into why the fix starts with the source policy document rather than the current application, why the power users have to be in the room from the first pilot, why procurement belongs at ideation and not at the end, and why speed, from OTAs to fast pilots, is never the same thing as the outcome you actually needed.
Timestamps:
- 00:01 — Why the US is now leading on AI uptake while process and policy slow everyone down
- 00:04 — Context is the ceiling: why the smallest gap derails an AI tool
- 00:11 — Blueprint for Government, source policy, and paving the cow path
- 00:13 — Design-time AI vs. token-maxing runtime AI
- 00:16 — Local models, warrants, and where a contracting officer should focus
- 00:24 — Dropping point solutions on the front porch of IT
- 00:25 — Getting procurement, CISO, and ISOs in the room to go idea-to-live in 90 days
Resources:
Doug Averill’s LinkedIn profile
Chelsea Roberts’ LinkedIn profile
Other resources: