GSA OASIS+

[formerfed]

More multiple award governmentwide IDIQ contracts.  GSA just issued the RFP for OASIS+, a pool of six multiple award contracts covering non-IT services with a variety of set-asides as well as unrestricted awards.  Their pre solicitation work apparently showed a large amount of interest from industry including companies that haven’t done business with the government before and small startups.  To allow industry time to digest the solicitation and award process as well as answer questions, proposals aren’t due until September 13.  A key part of the proposal submission is industry self-evaluation and scoring which they adopted from their IT GWAC contracts.

I won’t be surprised if OASIS+ exceeds the number of CIO-SP4 protests.  The entire concept is just a carryover from the IT GWACS and pushed further by the governments latest trend to jump on the private sector success with category management.  It’s also something many inexperienced 1102s like because all they have to do for a procurement need to have a program office write up a statement of work and maybe evaluation criteria if they don’t do LPTA, post a fair opportunity notice, and let program offices do an evaluation. 

[Voyager]

20 hours ago, formerfed said:

The entire concept is just a carryover from the IT GWACS and pushed further by the governments latest trend to jump on the private sector success with category management.  It’s also something many inexperienced 1102s like because all they have to do for a procurement need to have a program office write up a statement of work and maybe evaluation criteria if they don’t do LPTA, post a fair opportunity notice, and let program offices do an evaluation.

Does anyone know firsthand at their agency if these 1102s are learning anything about contract formation in doing this?  For example, offer-and-acceptance, consideration, negotiation basics, etc.?  I am just curious whether or not WIFCONsinites think these jobs are better off being replaced by artificial intelligence.  A darn good reason for that IMHO would be if the duties do not require the doer to gain knowledge, skills, and abilities in any important topics to accomplish the tasks. 

[C Culham]

57 minutes ago, Voyager said:

Does anyone know firsthand at their agency if these 1102s are learning anything about contract formation in doing this?  For example, offer-and-acceptance, consideration, negotiation basics, etc.?  I am just curious whether or not WIFCONsinites think these jobs are better off being replaced by artificial intelligence.  A darn good reason for that IMHO would be if the duties do not require the doer to gain knowledge, skills, and abilities in any important topics to accomplish the tasks. 

AI might just work.  Yet my mind runs wild on such things and my first thought is that maybe, just maybe AI ought to be used to figure out if one (government and its contractors) should get to OASIS via the guiding principles of FAR subpart 8.001 through 8.005.   

[Voyager]

2 hours ago, C Culham said:

government and its contractors) should get to OASIS via the guiding principles of FAR subpart 8.001 through 8.005

A basic reading of these FAR Sections tells me you think AI ought to direct activities with service requirements to the following sources, in order of priority.

1. Procurement List maintained by the Committee for Purchase From People Who Are Blind or Severely Disabled;
2. Federal Prison Industries, Inc.;
3a. Federal Supply Schedules, 3b. Governmentwide acquisition contracts (GWACs) - like CIO-SP4, 3c. multi-agency contracts - like OASIS, or 3d. blanket purchase agreements (BPAs) under FSS contracts.

Is this what you mean? 

[C Culham]

16 hours ago, Voyager said:

A basic reading of these FAR Sections tells me you think AI ought to direct activities with service requirements to the following sources, in order of priority.

1. Procurement List maintained by the Committee for Purchase From People Who Are Blind or Severely Disabled;
2. Federal Prison Industries, Inc.;
3a. Federal Supply Schedules, 3b. Governmentwide acquisition contracts (GWACs) - like CIO-SP4, 3c. multi-agency contracts - like OASIS, or 3d. blanket purchase agreements (BPAs) under FSS contracts.

Is this what you mean? 

Is there not the decision process of the "Sections" regarding what source before accomplishing a task (procurement) associated with something as specific as using OASIS?  A decision required of the FAR sections?  If yes, then yes that is what I mean.

Additionally I misplaced my reference to "contractor" as by my read there is also a decision required by Section 8.005, as appropriate, correct?   If so use of AI to make it?

Just trying to sort out the thought of use of AI in Federal acquisition and this instant thread.  A decision matrix tool or just a tool to assemble the appropriate solicitation package when a decision of where should I buy it from is made?

[Voyager]

While these policies are helpful and do need a boost in their compliance (both from agencies and contractors), I am wondering if, once the selection of a purchase under a GWAC or the OASIS multi-agency contract (MAC) is the decided path forward, does anyone think their contracting offices should always employ a human mind to get from that decision to contract award?  Specifically what KSAs are these 1102s learning in doing so?

Genuinely curious, as I do not purchase under these currently.

[formerfed]

A human mind really needs to make those kind of decisions. So much also depends upon management philosophy, “politics,” and discretionary judgements.  Once while working with an agency wanting to award a pool of IDIQ contracts, I asked why don’t they award BPAs under GSA Schedules or use GWACS.  The reply was “their office gets more workload credit for awarding their own contracts!”  There are also factors like socioeconomic considerations and even management preferences.  

But I can see where AI can play a big part in both pre-award and post-award placing of orders under a concept like OASIS.  For example AI could help GSA is selecting discriminating factors in the various phases of source selection based upon experiences of others including government and industry.  It could help the private sector in preparing proposal responses and the government in evaluating.  In a similar manner it could assist ordering agencies in crafting order solicitations including order selection procedures as well as evaluation.   At a very basic level it could produce task order selection processes based on experiences of other uses for similar services.  For many orders I can envision little if any 1102 involvement.

[KeithB18]

As these GWACs expand the number of awardees, to what extent are they just recreating open market, full and open competition? An early benefit of these contracts was the relatively small number of vendors. You could legitimately sort through the short list and assess whether those vendors could fulfill your need. That seems less possible to do as the number of awardees grows. Would we expect better offers through a 100 vendor GWAC solicitation or through a 12.6 combined synopsis-solicitation posted at sam.gov?

I honestly don't know. The protest threshold for GWACs is still a major draw.

[C Culham]

23 hours ago, Voyager said:

Genuinely curious, 

I will say I am as well with admittedly a very minimum grasp of AI.   That said my additional thoughts....

A CO has to sign the award, or at least in most all cases, an "AI" can not do that.  So what level of confidence (review) does the CO put in the AI conclusion versus that if provided by an 1102?  Just say okay and run with the AI decision or ?   And if the "or" who does the CO talk to for clarification on why is the recommended awardee being suggested?   And then further what if the decision is protested and it was an "AI" determination for award document how does the CO defend it?   "It was done by AI so therefore it is good" seems like a hollow argument.

How about have the very same procurement conducted under OASIS done by AI and at the same time have an 1102 do it and compare the results.  Does doing so help sort out if the 1102 adds value?

I researched OASIS contract clauses, I find it interesting that 52.208-9 is not in the parent contracts.  Not necessarily right or wrong just interesting.

My thoughts go on and on with regard to any AI use in Federal procurement....I need to study more before I say anymore about my thoughts!

 

[formerfed]

4 hours ago, KeithB18 said:

As these GWACs expand the number of awardees, to what extent are they just recreating open market, full and open competition? An early benefit of these contracts was the relatively small number of vendors. You could legitimately sort through the short list and assess whether those vendors could fulfill your need. That seems less possible to do as the number of awardees grows. Would we expect better offers through a 100 vendor GWAC solicitation or through a 12.6 combined synopsis-solicitation posted at sam.gov?

My thoughts exactly.  An enormous level of resources will be devoted to this with a bottom line I don’t see as beneficial.  GSA has spent a couple years just getting to the current level.  I’m sure vendors will file pre-award protests before proposals are even submitted.  It may take a couple years to just get through clean awards.  Then ordering agencies must comply with fair opportunity among dozens and dozen of contractors before placing orders.  My question is why not conduct research among GSA Schedule contractors, select a handful of the best qualified, and do a quick competition probably ata fraction of the time and effort?

3 hours ago, C Culham said:

A CO has to sign the award, or at least in most all cases, an "AI" can not do that.  So what level of confidence (review) does the CO put in the AI conclusion versus that if provided by an 1102?  Just say okay and run with the AI decision or ?   And if the "or" who does the CO talk to for clarification on why is the recommended awardee being suggested?   And then further what if the decision is protested and it was an "AI" determination for award document how does the CO defend it?   "It was done by AI so therefore it is good" seems like a hollow argument.

How about have the very same procurement conducted under OASIS done by AI and at the same time have an 1102 do it and compare the results.  Does doing so help sort out if the 1102 adds value?

I would think in practice a program office PM or SME would review anything coming out of the AI process.  They would be able to communicate with the CO.  As far as the business side I think a CO would do a similar type review.  In any event unless somehow there is changes in law and regulation, the CO signing is the responsible official.  What I see though is contract specialist time greatly diminished or even eliminated.

[C Culham]

1 hour ago, formerfed said:

I would think in practice a program office PM or SME would review anything coming out of the AI process.  They would be able to communicate with the CO.  As far as the business side I think a CO would do a similar type review.  In any event unless somehow there is changes in law and regulation, the CO signing is the responsible official.  What I see though is contract specialist time greatly diminished or even eliminated.

I will admit that I have a limited view.   Million dollar task orders not, but 100 thousand task orders yes.   My most recent experience t dates back 15 years ago but from there reaches back almost 40 years.  I saw limited effort of PM/SME's with regard to any review with regard to the nuts and bolts of a procurement.  It was here you go now go buy it.   That is not to say that they were not reengaged in TEP's and review of TEP effort documentation but I honestly would say that it was cursory at best.  Heavy lifting of creating a solicitation and every thing to award, inclusive of evaluation criteria was left to the 1102/CO again with cursory review ("Will this work for you").

I am not saying 1102/CO time could be diminished but at what cost in all kinds of ways?   This will not be appreciated and again I realize I am speaking with minimal KSA's regarding AI but I liken it to the argument that performance based acquisition would be less costly in all kinds of ways.   I hope the implementation of AI has measurable analytics to prove that it is an actual benefit.   Articles I read at this point suggest the benefit in all kinds of ways is questionable.  

[formerfed]

1 hour ago, C Culham said:

I will admit that I have a limited view.   Million dollar task orders not, but 100 thousand task orders yes.   My most recent experience t dates back 15 years ago but from there reaches back almost 40 years.  I saw limited effort of PM/SME's with regard to any review with regard to the nuts and bolts of a procurement.  It was here you go now go buy it.   That is not to say that they were not reengaged in TEP's and review of TEP effort documentation but I honestly would say that it was cursory at best.  Heavy lifting of creating a solicitation and every thing to award, inclusive of evaluation criteria was left to the 1102/CO again with cursory review ("Will this work for you").

Unfortunately much of the current 1102 work I see doesn’t demonstrate that level of “heavy lifting” and “will this work for you.”  In prior years, it usually wasn’t like that.  Of course there are lots of exceptions where the 1102 does both the work of the contracting expert and the program/technical expert.  But for the most part where more complex services like those covered by OASIS+, 1102s generally pass on responsibility for creating SOW/PRS/SOO’s to program offices including preparing much of required supporting documents like SSPs, APs, justification, and other things.  Most clearly see a clear separation of duties and responsibilities.  I think in the past, people throughout agencies turned to 1102’s as the contracting experts and the expectations were 1102s take the lead on most things contract related.  Not do much now.

[REA'n Maker]

I would consider it a major victory if someone would develop a reliable clause logic engine- it's inherently a rules-based process, which should be extremely easy to automate.  Yet to date, all we've gotten are useless macros that miss the most obvious clauses imaginable and in general crank out pages and pages of garbage (e.g., not selecting any Part 36 clauses for an A-E procurement)  Having sat on the contractor side of the house for 20 years, I can affirmatively state that the people writing code for every Federal contract writing system don't have the slightest clue how these systems are to be used.  My point being that I don't see AI being a factor in Federal procurement anytime soon. 

[KeithB18]

On 6/22/2023 at 1:31 PM, formerfed said:

My thoughts exactly.  An enormous level of resources will be devoted to this with a bottom line I don’t see as beneficial.  GSA has spent a couple years just getting to the current level.  I’m sure vendors will file pre-award protests before proposals are even submitted.  It may take a couple years to just get through clean awards.  Then ordering agencies must comply with fair opportunity among dozens and dozen of contractors before placing orders.  My question is why not conduct research among GSA Schedule contractors, select a handful of the best qualified, and do a quick competition probably at a fraction of the time and effort?

 

Good point; I hadn't considered it from that angle. I wonder if anyone at GSA is considering the bigger picture: Is the squeeze worth the juice? 

A couple years ago I looked into how much activity the original OASIS contracts were getting and a surprising number of them had zero awards. So I even wonder if this effort is even worth it for the contractors. 

[Contracting_in_Wonderland]

On 6/23/2023 at 10:32 AM, REA'n Maker said:

I would consider it a major victory if someone would develop a reliable clause logic engine- it's inherently a rules-based process, which should be extremely easy to automate... 

Good lord, yes to this. Each time I hear of a new clause logic tool, I get excited and then immediately let down when I attempt to use it. I don't know why I'm so gullible after all this time. These darned rose-coloured glasses I tend to wear, I guess. 

[formerfed]

On 6/29/2023 at 9:35 AM, Contracting_in_Wonderland said:

Good lord, yes to this. Each time I hear of a new clause logic tool, I get excited and then immediately let down when I attempt to use it. I don't know why I'm so gullible after all this time. These darned rose-coloured glasses I tend to wear, I guess. 

The idea of an easy to use and accurate tool doesn’t seen complicated at first.  But once someone gets into it, it quickly gets complicated.

Take a look at this.  https://www.acquisition.gov/smart-matrix  Starting out it looks simple, but once we dig into some of the FAR prescriptions things get cloudy.  Often the tools require users to answer questions about the specific contract action underway or fill in the blank with required information.  That can be easy for some clauses but in other cases the issue gets complicated.  Simple Q&As don’t work well.  If the question isn’t phrased properly or subject to multiple interpretations, errors result.  Also a proper tool needs to cross check between clauses and situations to ensure conflicts don’t occur.  Plus some clause situations are exceptionally complicated - consider some of the lengthy debates here on interpretations of clause applicability.  FAR supplements as well as specific agency policies needs added if nothing more than to complete the FAR coverage like “in accordance with agency procedures.”  Finally there are continuous FAR updates.

All this can be solved by a diligent effort that developers and SMEs.  But those sources say the cost isn’t worth it because the government market doesn’t justify the expense.  So until something like AI advances where the cost is much smaller we are stuck with tools that just partially meet our needs.

[Voyager]

30 minutes ago, formerfed said:

the government market doesn’t justify the expense.  So until something like AI advances where the cost is much smaller we are stuck with tools that just partially meet our needs.

😰 You mean I have to learn what I'm buying??  And what the FAR drafters intended??

I kid, but really, only a human mind should perform this duty.  Think of the KSAs gained by talking to the PM during solicitation development, or by reading the legislative history on a clause.  Keep these research functions for the people; give away clerical duties to machines.  It will free up the time investment needed for research.

[C Culham]

4 hours ago, formerfed said:

But once someone gets into it, it quickly gets complicated.

That complication is increased by practitioners.  Just for the heck of it I got on SAM.gov and looked at the very first commercial item solicitation that came up.  To make it simple, if the very person(s) that created the solicitation advise and help create the AI tool adherence to the FAR explicit FAR guiding principles will be lost even more than now.

As I provided in another post but stated another way.  I suggest and fear there will be the day that AI discretion will be acceptable reasoning in the profession of acquisition.  But I guess if an AI can pass the new FAC-C exam then so be it.

[formerfed]

20 hours ago, C Culham said:

As I provided in another post but stated another way.  I suggest and fear there will be the day that AI discretion will be acceptable reasoning in the profession of acquisition.  But I guess if an AI can pass the new FAC-C exam then so be it.

That machine based “expert” reasoning has been around awhile.  25 years ago I used to challenge some very experienced contract specialists on why some clauses were erroneously included in documents.  The common response was like “I don’t know…the system put it in.” 😀

[formerfed]

On 6/30/2023 at 11:56 AM, Voyager said:

 

I kid, but really, only a human mind should perform this duty.  Think of the KSAs gained by talking to the PM during solicitation development, or by reading the legislative history on a clause.  Keep these research functions for the people; give away clerical duties to machines.  It will free up the time investment needed for research.

I agree.  But think down the road where the best contracting SMEs “do the teaching” for AI.  Then AI canvasses all the best practices and lessons learned including success stories, failures, protests, etc., and the reasons.  AI might end up producing better acquisitions than the bulk of 1102s.

[Contracting_in_Wonderland]

On 6/30/2023 at 11:19 AM, formerfed said:

The idea of an easy to use and accurate tool doesn’t seen complicated at first.  But once someone gets into it, it quickly gets complicated.

...

All this can be solved by a diligent effort that developers and SMEs.  But those sources say the cost isn’t worth it because the government market doesn’t justify the expense.  So until something like AI advances where the cost is much smaller we are stuck with tools that just partially meet our needs.

I don't disagree. It's just frustrating when some of the tools do not even capture clauses that are prescribed as "in all solicitations and contracts." I get the ones that require a little more thought, but when they can't even get what I would call the easy ones right I just wonder why someone even bothered to build the tool at all! I would expect this to always be a hybrid human-AI activity anyway because of some of those tricky ones, but I'm still hopeful that one day there will be a tool that gives a good jumping off point. 

I'm not even an active CO anymore, but am still holding onto the hope for if/when I'm writing contracts again.