AI & The Future of Grant Funding

Discover how AI is revolutionizing the grant funding landscape, empowering organizations to streamline application processes, enhance decision-making, and improve the overall efficiency of grant programs. This fireside chat will be particularly of interest to grantseekers and grantors alike.

In This 1-Hour Session, We Covered:

  • How to evaluate AI solutions in grantseeking
  • How to leverage AI to secure government funding
  • How grantors may leverage AI to ‘fix the form”
  • What does grant writing look like in ten years
  • What does grant making look like in ten years

About the Speakers

Sedale Turbovsky, CEO & Co-Founder, OpenGrants

Sedale Turbovsky is the CEO and co-founder of OpenGrants, his fourth venture-backed startup and second in the public sector, which aims to transform the grant funding landscape by making it more accessible, equitable, and transparent. Sedale firmly believes that technology can be used as a powerful tool to democratize access to resources and opportunities, and is passionate about improving government and the citizen experience.

Prior to embarking on his entrepreneurial journey, Sedale worked as an issues management consultant supporting firms working with government and in heavily regulated environments. He has a long history of leading successful teams, raising capital, and building startups. Apart from his work at OpenGrants, Sedale is an investor and advisor to a variety of startups.

Dimitar (Mitko) Simeonov, CEO, Pioneer

Mitko is the CEO and co-founder of Pioneer, his third early stage startup. He has a decade of experience in startups and before that built AI language models during his master thesis at MIT. Pioneer is an AI-powered consultancy, whose mission is to coordinate the funding for rapid decarbonization, and it achieves it by helping companies identify and qualify awards, apply effectively, and manage the process of interfacing government agencies. Pioneer was named after its customers – innovative technological companies solving the biggest problems of the day, and is actively hiring.

Read the Transcription

Please note, this transcription is automatically generated and may contain some spelling and contextual errors.

Sedale Turbovsky: All right folks, welcome in. Super excited to have y’all here. We are going to be talking about AI and the future of grant funding. So I’m just going to go ahead and throw this up here. Now, if you’re just joining us just a little bit of housekeeping, you’re in the right spot. This is the OpenGrants webinar on AI and the future of grant funding and just a little bit of housekeeping items as y’all trickle in.

We’re going to give folks a few minutes to come through. So those who are like opening up the emails and clicking on the link and getting in the door. In the meantime, our show today, our webinar today is going to start off with a bit of a fireside chat. And then we’re going to move to a question and answer period for y’all.

We do have the chat itself shut for these webinars for everyone’s safety and comfort, but you can use the QA tool. And please do submit answers through the QA for myself or esteemed guest here, Mitko. So we’re super excited to have an expert on AI, who’s building one of the coolest companies, I think in this space.

And we’ll get into that in a minute. But yeah, just wanted to welcome everyone here. My name is Sedale Turbovsky. I’m the CEO and co founder here at OpenGrants. And this webinar is called AI and the Future of Grant Funding. We’re going to talk a bit about AI and how it’s impacting this space. And so just a little bit of background.

Myself and the team here at OpenGrants, we specialize in doing a couple of things. We connect folks to grant funding and we connect folks to technical assistance. And so on the OpenGrants platform, you can find both grants as well as those folks who can support you, specialists who can support you in your journey to secure grant funding, write grants.

Applications, those kinds of things. So super excited to have you all here. Thanks again for joining. Once again, my name is Sadal Trubosky. I’m the CEO and co founder here at OpenGrants, and I’m super excited to be joined by Mitko. He is the founder, are you CTO, CEO? Anyways, he is one of the founders of Pioneer, which is a super cool company.

I’m going to let him introduce himself a little bit and a little bit of his background.

Mitko Simeonov: Thank you. So now thanks for the introduction and thank you everyone for joining us today. So today I was like a great friend and supporter of Pioneer has been for a while. I’m really excited to be chatting with you about some of the stuff we’ve been talking, separately, but now in a broader forum.

So yeah, my name is Miko and I’m the founder and CEO of Pioneer and at Pioneer we are early stage company and our mission is to coordinate the funding for rapid capitalization of our economy. And we do that, that by helping companies apply for. non dilutive funding such as government grants and contracts, but also non governmental sources.

So I also have extensive background in AI and software development and so happy to talk about how basically what’s happening right now in the AI space and then how does that relate to grants and other non dilutive applications.

Sedale Turbovsky: Awesome. Yeah. And thank you so much for joining us. I, you’ve been studying AI literally for years like tens of years.

And I, one of the questions I had that I was just thinking and thinking a lot about is, at least for me, someone who is not super technical. This whole thing, there’s AI was always out there, but then all of a sudden, it seemed like it was like, there was just been this explosion with generative AI.

And I just wanted to get your perspective on, like, where we are in terms of technology development, because, we see this. Technologies progress over time that become more ubiquitous and then they get adopted by everybody. And there’s that, there’s that classic adoption curve that we’ve all seen.

Where do you think we are in that curve as it relates to AI? Is this is this a special moment we’re experiencing or is this sort of just like business as usual as technology develops and evolves?

Mitko Simeonov: It’s a fantastic question. So I’ve been, I just wanted to just share a little bit about my background about so that people can.

know how much to trust my opinion about AI, but I’ve been I did my master’s in MIT more than 10 years ago. And at that point I was doing natural language for robot manipulation of objects. And at that point, neural network were considered like a failed approach. And later on, deep learning came up and Recurring neural networks and now the transformer architecture.

So I also was a Twitter at some of their data science machine learning team where I helped build what is known now as the algorithm. So I’ve had, I had some practical experience as well. There’s obviously people like Andrej Karpali that are way more out there. So I wouldn’t say I’m the biggest expert, but I have enough to share some, interesting thoughts and to a question like, is this business as usual?

So I’d like to think about it. Is this an incremental? AI improvement, or is there actually like a big step function here? And one of the very exciting and interesting architectures for AI right now is transformers. And even before we go into that, I wanted to talk about AI, the phrase AI has been around for decades.

It means many different things. So when I talk about AI today, I’ll specifically talk about large language models or LLMs. I’m going to use these two interchangeably. And then I might say just traditional AI or classical AI when I’m referring to the broader. And so in terms of… LLMs they’re mainly based on the transformer architecture.

And what is very exciting about this reformer architecture is that it’s much more like a computer than a calculator. The previous, a lot of the previous models are very optimized for a specific purpose. So it’s like doing a calculator, it can do addition, subtraction, certain statistical pattern matching, whereas transformers, they act much more like a computer that can iterate.

They can do. More complex tasks, and we’re seeing that with GPT 4. And so they have much more expressive power. And actually, theoretically, this is well understood in tier of computation, they are Turing complete. And that’s, it’s very exciting to me. That actually made me a lot more excited about what’s going on.

Because right now, if we think of LLMs, we can think of them as a programming language that is now accessible to people that are not just software engineers, anybody who knows English language and can be precise in what they want to achieve. And also the transformer architecture has been around for close to, I think, six years plus.

So it’s been very exciting to see that and follow that, but it hasn’t been accessible. Today we are at an inflection point starting sometime last fall when ChargeGPT launched, and that became the price point and accessibility and user interface became much more easy to access to broad set of people.

So now a lot of people are starting to learn about prompt engineering and other techniques. So I would say yes, there is a specific inflection point and I can talk more about it later, but it’s, it becomes a great platform for innovation.

Sedale Turbovsky: Awesome. No, this is super exciting. It’s definitely, something we’ve experienced here.

When we launched OpenGrants, we had these ideas about how we were going to transform how people write grants and access grant funding and frequently we just came up against the fact that it was a little cost prohibitive to do some of the things that you can do like on a weekend now. Yeah there’s definitely, we’ve seen a change here and it’s interesting to hear that from you as well.

I thought one of the things we could start out with today before we dive into the specifics of grants is just like high level, since we have you here high level, there’s a lot of orgs. I’m sure there’s startups and other groups on the webinar today who are probably interested in integrating these tools.

And I just wanted to get your thoughts, as everyone’s thinking about AI and like maybe getting a little FOMO are we being left behind or should we even use this? And we’ll get into that later. What are some thoughts that you have just practical advice for folks who are looking to integrate AI solutions into their workflows or into their product offering any like pointers you could give folks who are thinking about like, all right how do I take advantage of all this cool stuff?

Mitko Simeonov: Yeah, so great question. So I’ll just break that into two parts. One is if you’re trying to use some kind of tool and want to bring in an AI tool, I would just say, just think about any other tool and evaluate on those dimensions. AI is like I said, it’s like a programming language. So if a tool is built with a better programming language, why would you care as a customer?

At the end of the day, you care about, does it solve your problem? Is it easy to use? Is it what is the user experience? What is the price? All those other parameters. And there is an interesting point with, to be made with AI power tools is A lot of them are thin wrappers around ChatGPT or some equivalent.

System with a better user experience. One of the things to be very careful, I would say, just be very careful about is understanding the data privacy protection, reading the terms of service, because one of the dangers is if you’re using any kind of proprietary company data there you want to make sure that it’s not going to be used to train the generic model that now everybody can use.

This is just as good as like publishing your data out on social media in terms of… For legal privacy, and this is something we’ve been taking very carefully, pioneering, creating a rigorous customer data protection policy with our customers. And from the other part of the question, if you’re a company that’s looking to and you have a product and you’re looking to integrate elements into your solution, you probably already have looked at it and done something about it already.

But, yeah, I would say making sure that the data privacy as well and then looking at how can you scale the capabilities and integrate and how can you be actually. Innovative there. I think that’s a really big challenge because it’s often there’s a bit of a design, you design a product around specific use case and maybe just thinking about, maybe I can ask you like, what do you think the future is going to be, but it’s going to be just to just think about the future and.

What happens when we have products that are just completely inspired by the use cases and the possibilities of AI? I don’t think we see that many product out there. Often they are something very still to you, but I’m excited to see something that’s like completely AI, LLM native still.

Sedale Turbovsky: Yeah, that’s awesome. I love it. Great advice. I think, just pragmatically approaching these things. And it is at the end of the day, a tool and should be evaluated as such. Great advice. I think what we’ll do let’s get into let’s get into grants and the future of grant funding.

We’ll, we will, Maybe take a step back and circle back to more of AI and the future and some of your outlooks there. But I’m super excited to talk about, first high level. You know what pioneer is and what y’all are doing. And then talk a little bit about just like how you see this impacting the future of grant making.

But can you give us, high level? What’s pioneer? Why is it exciting? Or why should people be excited about it?

Mitko Simeonov: Yeah, like I said before, pioneer mission is to coordinate the funding for rapid decriminalization of the economy. And we have companies identify, apply, and later on comply with. Various different types of awards and bids.

And we realized this is a big, messy process that requires a lot of project management, document preparation decision making in a group setting. A bunch of related to business development and sales. We want, we realized that if we just build like a standalone tool, that’s only going to sell like a very small subset of those problems.

And so the way we set up pioneer. It’s much more like a, from the point of view of a customer, it’s more like a tech company that is dressed up like a consultancy from the point of view of customer. And so we want to, that allows us to actually address the bigger challenges and keep on iterating, really solving the big, important problems that is on fire.

And one thing we realized especially with the grant making space and grant writing, there’s a bunch of like really amazing people, but there’s just not enough people who want or can do this type of work. And so we want to help scale the workforce and give. Really superpower super super powers to those who can or want to do it.

Sedale Turbovsky: I love that. Yeah, it’s, that’s definitely something, folks within our ecosystem struggle a lot with scaling technical assistance, which is what we call it. But the work of grant writing and strategic support, and this is one of the reasons. And, as as you alluded to earlier.

I’m, personally, huge fan of Pioneer. I’ve also contributed some resources in that direction, full disclosure. But the reason I’m so excited and have gotten involved with Pioneer is because of this. This is a, it’s a super cool point that you bring up of giving. These really smart talented folks who have developed this very niche kind of capability of operating in the, like the grants and government space superpowers is just so exciting to me because they always, I worked in that space for some time and saw how overworked and maxed out on bandwidth.

All these folks were who were building and doing these really cool things. And so it’s very exciting. Very cool. I’d love to talk a little bit more about just. If you have any examples of what you’ve all been into just to give just to give folks more of a concrete taste of what Pioneer does.

Mitko Simeonov: Yeah. One example is I think one of the first thing we built internally that was actually, like I said, it’s also thin wrapper around all the LLM models is draft generator. So we work with our customers that we work with, usually those are companies series a, but we also can work with publicly traded companies.

To help them apply for awards and usually what we do is we bring some of their documents and contents and we’re able to use that as a base for enough context so that our language models can use that. And one of the first things we did is first draft generator, right? And so that allows people who are doing grant writings, instead of spending a lot of time on figuring out the wordsmith, All the texts, they can get a really great starting point and that can save significant amount of time, specifically in the writing stage.

There’s a bunch of other projects that we have underway. But things like processing the funding opportunities, announcements of the RFPs and extracting the requirements, also matching them with that body of content for each customer and figure out what are the most best qualified opportunities generating the right strategy and also.

It is a big coordination challenge. So you want to coordinate in the usually what you might have seen that sometimes the writing of the grant is a bunch of the work, but it’s not necessarily the hardest work. Sometimes it is. Pulling out the information out of the applicants and getting them a buy in about the approach you want to do because at the end of the day, it is, they’re the final decision maker who owns the application.

They’re name is signed on the deadline. So they have, you have to get their buy in. And I think sometimes that involves actually multiple people on their end. So a lot of coordination. And that’s why as a part of our mission, the first word is coordinating. Or rapid the funding for rapid organization.

Sedale Turbovsky: Super exciting. I think, we’ll get into this a little bit later. But, some of the most important work that I saw, or that we engaged in as strategists and consultants in this space was often right, like grant writing is just this after the fact activity to like, communicate all this stuff you learned, especially if you’re a consultant, but it’s not.

It’s not really the core activity. It’s just a byproduct of the activity that you did and that’s super exciting. I want to encourage the audience because I just saw 1 come in. Please do feel free. We will open it up for a Q and a, but feel free to use the Q and a tool while we’re speaking to send us any questions.

You might have would love to. I love to address those. So please do chime in. I want to get into more of this, like, all right, what does this mean for the future? So we’ve talked about AI at large. We’ve talked a bit about what Pioneer is and one of the reasons I’m super excited about it is I do think You know, it’s this kind of thing is the key to accelerating deployment of capital for underserved communities for climate for many other things across the board, just because one of the core constraints that’s been very visible in the industry is a lack of scalable TA and really like those service providers who can help you strategize and do all the things you’re talking about, I would love to get your take on this because, it’s something I think it’s an uncomfortable thing that sometimes we, we struggle with when we talk about like the future of this is some people feel very protective about their roles.

Other people are more, open to embracing this but this really has been this work has been the domain of like specific subject matter experts for a long time. There’s a whole cottage industry built around it in consultants wise. How do you think about the future of this space?

And how do you see the role of those subject matter experts and consultants involved evolving as technology like this develops?

Mitko Simeonov: Yeah. Like I said, we want to empower them with superpowers. We have a Internally, we call it the iron man, iron woman suit that we’re building or like a formula one race car.

So we want to actually make that job a lot more fun and also bring that to the high level of strategy, project management, reducing a lot of the cognitive load for the applicant. And making their job become now the trusted source of information, right? Because at the end of the day, the final customer will probably not want to fully trust any kind of software tool.

They want to always have a person they want to talk And a grant consultant, the grant writer, their job would be keep the final customer applicant accountable for all the things they have to do. But also act as that this is a big waiting source of information for them. Just like I might be asking for some high level legal advice and explain to me how certain things work, but at the end of the day, I might want to always just ask my lawyers for when I have to do something with high stakes and grants and contract application is incredibly high stakes.

I’ll be actually excited for increasing and elevating the status. Of this job and profession because we think we can actually increase the value that each of those people bring and then that will come with increase in status and earning potential.

Sedale Turbovsky: Awesome. Yeah. No, I think that that’s great I wonder if you could talk through any you know I think there’s a lot of opportunities for people that you know, we talked a little bit about just You know, the ubiquity of this technology and the accessibility.

And I’d love, I’m happy to give one example. We’ve worked on some pretty cool things. If you want to test this out, if you’re an entrepreneur and just see where like the bits and pieces are like, you can go and. Access GPT or cloud or any of them, many of the models for free.

And you can write queries and, ask them questions. And. I just wonder as you see, like these models develop and different applications for the industry. Do you think that the, do you have any pointers for like consultants who are trying to like get into this space or for founders who are just like, want to go out and explore any thoughts on What model to use over a different one.

I know you’ve been, and maybe you don’t have any thoughts on that, but I know you’ve been looking at this space for a while. Any thoughts on which ones are performing best or where’s the best place to like, just start your search and start experimenting?

Mitko Simeonov: Yeah, definitely. I would say at the moment, GPT 4.

Is the best model. And what makes it the best is the ability to actually follow instructions much more accurately than previous generations. And I think that’s very important. And then actually, that’s the main one we are using internally for a lot of our systems. And that allows us to have much higher quality of output with a lot less engineering effort put behind it.

I want to flag one of the fundamental limitations. For a lot of those models is the context length, namely how much text you can put additional context for the model to know what you’re talking about. And usually that’s like a few thousand words, so it’s not a huge amount. I think that’s really where those models are really trained to have the attention within that period.

And trying to extend that beyond working with an infinite, quote, unquote, infinite amount of knowledge, or just unbounded amount of knowledge, that is one of the challenges that we’re also trying to work towards. In terms of, so other models, Cloud specifically, they have the 100, 000 token context, which actually is significantly larger than, GPT 4, but I haven’t actually extensively, super extensively tested it, but the understanding is it’s not at that full level as well.

So I would consider using Cloud for larger if you have to pull in a lot more of the FOA text in there and using GPT 4 in other cases.

Sedale Turbovsky: Yeah. Awesome. Great insights. This is really cool.

Mitko Simeonov: I would want to just still flag the ability to like making sure that the data is not going to be used by OpenAI to train their models because you might be working with very sensitive IP data from customers.

Sedale Turbovsky: Yeah. And there’s some cool innovations happening around, retrieval augmented generation and other approaches to help you, keep some of that data in place, but yeah, all good considerations for folks looking to dive into this space. I think one of the things that I think is really cool, OpenGrants has this thesis of, we think one, like obviously grant writing is this, it’s a, it’s Communications exercise, right?

You’re taking all this info that you’ve derived from either your company or other places, and you’re synthesizing it into a document, and you’re like sending it to a funder. And, 1 of the things that AI does really well is, helping create those documents. And then also there’s the possibility and capability on the other side of the table to use AI to understand those.

Pieces of information. Our thesis is like very soon. We don’t think that people will be writing grants as much as you’ll just be like transmitting data. And we always encourage folks to get really good at storytelling to understand, the reason they’re telling the stories and how the the impacts and implications of the data and impacts they’re having relate to funders.

Because at the end of the day, you’re just trying to find funders who. Have the same goals as you, right? Whether that be climate or cancer research or building new technology for warfighters. It’s all there’s people out there who are trying to fund it, and I’d love to get your thoughts on, the role that you see Pioneer playing, but also just overall, your reaction to that.

Do you think that’s possible in the next couple of years? Where do you think we, where do you think we’re headed when it comes to this technology andSo, As it relates to specifically the the act of communicating for grant funding.

Mitko Simeonov: Yeah, first of all, I love that vision. I’ll be skeptical if I think it’s going to happen in a couple of years.

I would say maybe in a decade, the chances are bigger, just the way, the speed with which government is moving. I think, yeah, a lot of the communication could be a lot more simplified, right? It’s still a useful exercise if you’re applying for grants to get your narrative straight, like you said. But I think that can happen in a much more efficient way.

One thing that we’ve noticed and heard is the way that actually the government grades those applications. Is already very robotic, right? So they hire sometimes volunteers, and they tell them these are exactly the steps they have to follow. And even when you write response letters, you have to use very robotic style.

So they leave almost no room for creativity. And that’s actually a task that is very well handled by a lot of those language models. And so I would hope that the government actually improves that and while maintaining fairness However, we didn’t want to start by selling to the government about that.

So that’s why we started with the companies. And then as a part of that, we’ve been trying to figure out what is the way to score the applications and the drafts and the package so far. So that is something that is going to evolve on our end. But also wanted to flag like one common misconceptions people might have about.

a lot of those like AI models that sometimes tend to make up stuff. And so some people thought oh, if we use those models to help us write the applications, is there going to be a flood of low quality applications? And I think that is actually very solvable and actually a lot of those models and tools, they help companies express and communicate their stuff much more effectively, right?

So they already have the base content, the research, the product that they built. And so this is more of a communication tool. Like when I was at one of my first job they wanted to create a patent. So I spoke to a patent lawyer for a little more than an hour. And they came back a few weeks later with 15, 20 pages of.

Patent style application, which I wouldn’t know half of the words there, but those are the right words into using that context. We’re hoping that a lot of that application will be something like that. And even if we dare to dream further, what if we actually flip the whole communication process on its head?

And instead of, applicants today applying for grants, what if the grants and funding sources apply directly to the applicant to say, We think you might be good. It might be a good fit for us to give you money for that. And that happens a little bit to some degree where some companies can get invited.

To apply. But we think that might be like even an interesting thought to consider.

Sedale Turbovsky: Yeah, no, we love that. Our bit of our mission has always been focused on making this whole ecosystem more accessible and equitable. I do want there’s some really good questions in here. So I would love to dive into some of these.

Mitko Simeonov: Yeah, I just wanted to maybe to say I would expect that even though it becomes a more efficient communication exercise by more efficient on the human level and interfacing with it, but I would actually imagine the packages, the application package to actually become much more comprehensive. And even larger I think it’s just a pattern we’ve seen with anytime we can optimize the process that process tends to become bigger as well.

Sedale Turbovsky: 100%. I love there’s a good one. There’s a really good question here. And I think people will be very interested in this. There’s a question. Can we get, can You get more info in can you get more into using chat GPT for writing grants? And yeah, I think, yeah, this is there’s a lot of opportunities.

We touched on this briefly and I’d love to get your thoughts. Let me go. But one of the things that’s been really exciting is just the rise of first of all, you can go and get you can sign up for open a eyes GPT. You can get the plus version for 20 bucks a month.

It’s super cheap. And you can do a lot of really cool things there. I’ve actually done some experiments where I just built prompts to say, Hey, you’re a PhD researcher. Applying for an NIH grant, respond to the questions, and then I just respond, and then I put the questions in from the grant, and, it wrote everything from a data management plan to you know like a research, a research proposal.

There’s a lot of ways to leverage the technology out there to do these cool things. Obviously, there’s limitations, as you mentioned. But I think this is Yeah, this is a great opportunity to it depends on where you’re starting. I think often, if you know how to use a tool or not, if you’re very familiar with this process, I think, some of these basic tools can be really useful.

But, we’d love to get your thoughts on like, how can, say you’re someone on this call, who’s like just really interested in like using some of these models. To start writing grants or to improve your workflows, any thoughts on like best places to start just to get your You know, dip your toes in the water and then learn more.

Mitko Simeonov: Yeah. Excellent question. So actually this is one of the first thing we started with before even like doing much of the product, we were actually using tragedy directly. Obviously there’s like some settings you have to configure there about history and using data for training purposes. But once you do that and even help some of our customers around the grind directly using that process.

And what I found myself. Doing was instead of actually worried about the actual word smoothing and fitting, but in word limits, I was able to think much more about what is the narrative, what are the main points they’re trying to make and then edit basically effectively start by editing the application on that level.

So I would encourage everyone to just start using it right away. I think it’s going to save you a lot of time. It’s available. It’s. It’s effectively free. If you get the plus version, you have access to really really solid models is going to save you a lot of time. So I think it’s a no brainer if you’re in the space to start using that as a base system and understand what are the capabilities and there’s certain things that it’s fails right now, unless you’re very good at prompting because.

It can have a bit of more of a politician style answer, which says a bunch of words that don’t mean anything. So I have to say, be concise, be very factual, and it’s still not going to do always a great job, especially if it doesn’t have the right content in its context to know about what they actually need to say.

So that is one of the challenges that we find with some of the answers there, that they can be a bit more high level, blah, blah, blah responses instead of very factual, precise, concise. Which actually helps with application quality.

Sedale Turbovsky: Yeah, no, that’s great. I think and we’re getting some great kind of follow on questions.

And I just want to bring up, one experiment that we did at OpenGrants is we used Zapier recently dropped a new tool called interfaces that uses I believe it’s retrieval augmented generation where basically, you can say, all right, I want you to use the model GPT to do the math, but I want you to pull from like a specific data set that I’ve curated for you.

And then you can build a, you can build a model to start to talk specifically maybe to your subject matter expertise. And that’s a little, you have to obviously pay to use that tool, but I think there’s so many cool opportunities and the more. The more I see and play with the different tools that are out there the better idea I get about okay, this might be useful for this kind of job.

But I definitely am starting to develop an idea of oh, some of these like tools need to be very purpose built for very specific things. Whereas for like general stuff, maybe GPT by itself is great.

Mitko Simeonov: Yeah, and I think we are building this kind of purpose to a systems around it.

Like I said, it’s up here. There’s like a lot of systems to get right specifically optimized for his mind.

I think it’s interesting following a question about I think one asked about. How can you adjust the style of JGPT to his own narrative style?

If you want to paraphrase this, I’m a big fan of JGPT and with my grunts, I have a thread where I have uploaded. Language from previous grants are old. And can you share some tips to make it more effective to capture and keep my content tone language? And is ChatGPT 4 worth a 20 month investment?

Yes. The last question, yes, it’s definitely worth it, especially that’s helps you be more effective at your job. And then tips for being more effective in capturing your own content tone and language. I would say that still depends on the amount of context that you can say. It’s really a lot of the stuff becomes about prompt engineering, and that becomes a challenge.

One thing that we’re developing like new techniques around that, but you have to then explicitly tell the GPT, this is my style. And I’ll answer this thing in this style, maybe explain more about what this style is. So you have to. One mental model I have about tragedy is like, it’s a really good fast intern, but it’s not like a super strong specialist.

So it’s an intern knows a lot about stuff, but you have to explain very clearly what you want, and I think that’s what we’re seeing with prompt engineering. A lot of that a lot of techniques there are coming up.

Sedale Turbovsky: Awesome. Now that’s, that is a great insight.

Mitko Simeonov: Being explicit rather than implicit is always helpful.

Sedale Turbovsky: Yeah, definitely. Melinda, you asked which one, please text name. I think you might have been asking about the different models. If you can just maybe throw Melinda, if you could throw a follow on into the chat to just clarify, we would love to address that. Let’s see.

Oh, there’s a grant strategy question here. So I’ll just answer this one real fast. I want to make use of foundation segmentation models based on visual transformers, fine tuned on public medical imaging data. How would I get started on a grant strategy? Great question. Honestly, folks like NSF and NIH would both be probably super interested.

There’s a lot of really great tools out there that could help you start matching the grants and building a funnels that you could then start applying for grants. I definitely sign up for the pioneer waitlist. They are super in demand these days. But if you go to OpenGrants, you can sign up for free there.

We’ll start matching it to grants. It’s all free. And we can get you connected to other folks. So yeah, just in terms of building a grant strategy around developing AI there’s a, quite a few public agencies that are interested in funding this kind of work right now. So definitely dive in and get started.

That’s the best way to do it. Get started, build that funnel, figure out, like, all right, here’s the universe, the possible start to understand which ones are going to be really relevant for you and then start applying. And you can use some of the tools and suggestions that we’ve outlined here to maybe streamline that process.

And I think there’s another great question here. That kind of dovetails nicely. That’s more on the side to question is how will I. Be able to support fundraising professionals with identifying relevant grant opportunities, i. e. Doing that prospecting and building that funnel.

This is a great question. We certainly have some thoughts and some things here at OpenGrants, but I don’t know if you want to throw anything out there about how y’all are thinking about this, but I can definitely chat about this as well.

Mitko Simeonov: Yeah, I think it’s becoming like an area of overlap.

And I think this is where we work with OpenGrants before. Using OpenGrants is API to help us identify what is, help us with the search process. But really, if you think about it, what is a great way to qualify if a certain grant or other just fundraising opportunity is relevant. And there’s a bunch of factors.

Some of the factors like what is the general information of the company, but also what is available database. Grants and OpenGrants have an amazing database. So we’ve been working with that. But then you still have to like filter through that and match in a very deep level each of the possible opportunities.

And sometimes you have to look at the eligibility requirements. You have to look at Also, what are the company priorities and timeline, internal timeline, what else is going to be there. So that’s where we see a lot of those flexible language model allow us to actually program our way into saying, being able to evaluate and look through much more deeply to a lot of various opportunities.

And I know OpenGrinds has been building something to help narrow that process as well.

Sedale Turbovsky: Yeah, no that’s great. And then, so a little bit more on this. Molly, it’s a great question. 1 of the ways that we’ve been approaching this is we used vectorized. We use vectorized data. And so we’ll look at the text kind of content of somebody who’s saying, hey, I’m doing a project, for example, that’s direct carbon capture.

And they’ll, we’ll have them submit a few paragraphs, two or three about their project, what they’re building, why they’re building it. And then we’ll include other data like location and what kind of entity they are. So there’s basic filtering that can be done, and this is, that’s really just the job of software, right?

Is, okay, are you a non profit? Then you’re, eligible for these grants that are also for non profits. What we try to do, though, is think a little bit beyond that, because we do. We think grant funding should be about the outcomes and not about like your tax I. D. So there’s different like tools we can suggest.

But the core of our matching is really around the specialized embedding. So we’ll say, all right the text of this is super similar to the text of this grant for direct. Carbon capture. And so that’s how you know what we show to our users at OpenGrants is if you type in a profile and create a profile and we run that against our database will say, all right this is a 98% match.

This is a 70% match. And so it is really useful in terms of building that initial funnel of the possible and then starting to evaluate. And I think the next step and what we’re really excited about is also. Especially for programs with a lot of history is looking at all the historical winners of those programs and then comparing your project to those who have been historically awarded and giving you a confidence score.

And so I think that’s, that’s a feature that I think will be forthcoming. A capability that we’re very interested in and yeah, so I, at a high level for fundraisers, it’s just like really streamlining all of the work that you’re doing and then, down the road, I think it could even help you with evaluating individual opportunities.

Let’s see some more questions. Oh, great question from Justin. We’re going to get into this a little bit later, but I’d love to touch on it now. And I’ll throw this one to you first, Mikko, and then I’ll follow up. What comments do you have about the ethical use of AI tools for grant writing, i. e. using chat GPT and writing grants, meanwhile being mindful of data sets and confidential information?

Mitko Simeonov: Yeah, I can mention a few times just being mindful about. Is the data leaking to any subprocessor? Very important to because that’s being ethical to your customer to the applicant, right?

There is also the question of being ethical towards The system in general and to those agencies and to what degree you need to disclose whether you’re using certain tools and not to help you with and people have been using a lot of tools to help them with the writing process. And I think what would be, I would consider analytical is just putting the output of the AMR models vanilla without any.

Post processing and being verified by human. I think that would be probably unethical because there might be what people call hallucinations in the model. Those are becoming more and more rare, but you want to always double check the human and sign off as a human that this is, you’re standing behind.

Every word there. So I would say just be careful about that aspect and be careful not to misrepresent any information that you actually have, but being using AI as a tool to just more effectively communicate the actual information and so that, so I would say that is one of the ethical concerns that we’ve heard about and yeah.

Actually, I’m bullish that applications written with the help of extra tools will be higher quality and easier for the agencies to actually process and evaluate. So I’m hoping we can actually make it a positive edX case there as well over time.

Sedale Turbovsky: Yeah, I agree with that. I think, one of the things we’re excited about on the sort of like ethics and equity side is, I’ll just give a concrete examples.

I have a friend who, quite dyslexic and has struggled his whole professional career to write emails. And now with AI, he can like, communicate much clearer and has a much easier time of doing that. And so I do think this will like, GPT and others will really streamline open up access for folks who are trying to apply for grants, but certainly, you want to be mindful, not only of, like, how the model arrived at its capabilities and what kind of things that ingested along the way but also, how you’re using that and representing those outputs to the world, as you mentioned.

Yeah, go ahead.

Mitko Simeonov: Yeah. Yeah. I just wanted to add especially around this kind of like area of justice and there’s like a related ethics question about more like a systemic question of who actually gets to win most of the grants and funding. And there’s been a bunch of stats about number of small businesses able to compete for grants has decreased.

And then the constant, there’s been like increasing concentration. We didn’t like, radius around the seat of highway bandits beltway bandits. We want to actually, we think there’s like a edX violation at the moment of, or just systemically the system is not really creating that justice and fairness.

And so we want to actually, one thing that I’m excited about those AI tools and all of you grant writers here in the audience, being able to use chat GPT and help more customers, more effectively that is actually going to increase. The fairness of representation and meritocracy around applicants and one thing will, yeah, I think if we can increase the actual merit of the applicants, then the system will be more effective and it will grow.

Sedale Turbovsky: That’s great. I love that vision. Just a quick pioneer question here. What industries and what industry sectors and subsectors have you had the most engagement with for your AI grant writing tools?

Mitko Simeonov: Great question. And thank you. So we’ve been focusing on climate tech because it’s part of our mission to decarbonize.

But even within that, we’ve been working with some battery tech company, logistics devices company and also EV companies. And right now we are quite interested in optimize our offering around. But yeah, all around just climate tech, innovative climate tech. That’s where we’re, that’s our idea side of customers.

And that’s where we’ll get most of the engagement. We’re less focused on developers who, create like large solar installations, stuff like that, because that’s a bit more deployment stage. We’re also just naturally excited to be helping people that are very innovative, very creative at least as early customers.

Sedale Turbovsky: Awesome. Do you have any recommendations for the best places to start UK, would you recommend ChatGPT or another option? I think I already know the answer, but go for it.

Mitko Simeonov: Yeah, I would say, yeah, ChatGPT as well. And I’m not, yeah, I think EU has a lot of, so we are a pioneer, we focus on the USA market, so I can’t talk in too much detail about the EU market, but.

And actually, UK is now, it’s now specifically EU after Brexit, but a lot of the grants in EU are now based on just English based language. So I think that makes it very much easier to search as well. Awesome.

Sedale Turbovsky: Let’s see another question here. How can I leverage AI for technical development pitches suitable for investors thoughts?

Mitko Simeonov: Yeah, one of the things we’ve been using just on the side, we always like to just experiment with new tools. There’s one called Beautiful AI, which you can just sign up and we’ve been using that for some of the deck presentations that we use sometimes when we have to communicate to our customers.

So that is one idea, one pointer but just, yeah you might also use it to help you outline and prepare the actual content.

Sedale Turbovsky: Yeah, that’s a good one. We’ve played and we’ve used for Dex and other stuff. We’ve also used Gamma and we’ve used Tome, and I think, there’s some others out there, but I would just say, a lot of these, once again, as we alluded to earlier, a lot of this is just like, these are tools.

I tell people all the time, A shovel’s a shovel. If you don’t know how to dig, then, it’s not going to help you, kind of thing, right? I do love there’s probably some tools that are better than others in this space, but I would just say find a tool and learn how to use it really well.

Even Gamma has been one that I’ve used a lot, and it’s great, and I keep using it because I know how to use it. And there might be better stuff, like Beautiful might be way better, but I don’t know. Yeah, definitely. I would say find a tool to use and make good use of it.

I think these other two questions don’t seem like questions, but if they are please clarify. Here’s

a good one. Are there AI tools out there that help identify government grants and contracts for disabled veteran businesses and then assist in the application process? Yeah . So this is a great question. I, I want to stress, at least for us at OpenGrants, that we are a vertical and technology agnostic.

If you happen to be a veteran-owned business or you happen to be just a small business, or you’re a technology startup, or you’re a nonprofit, any of those. We really do serve everybody. So you can come, you could create a profile for free. We’ll help identify the grants and then we’ll help we’ll help connect you to technical assistance and grant writers and folks who can help you actually build the application and submit.

And I believe Pioneer is as much the same, but Nico, feel free to clarify a bit.

Mitko Simeonov: Yeah, it’s not our focus right now. So we want to make sure that in the areas that we focus on, we can provide most value. So we haven’t actually done that. If one of our businesses that we work with happens to be that, then we will do an extra step and to look into that as well.

Sedale Turbovsky: Awesome. Very cool. And I think that dovetails there’s another question here is, is pioneer for government funding only. I believe that’s what you’re focused on solely, but we also

Mitko Simeonov: have been supporting customers with basically all their on the ground that we already helped customer when it was the Caltech rocket fund, which is non governmental and we’ve also helped companies apply for various corporates.

Applications as well.

Sedale Turbovsky: Awesome. Very cool. That’s really good to know and need to hear. I, we are closing in on the last 10 minutes here. And I did want to get into a bit of a discussion with you about, more about just the, the current kind of vibes around a I and get your thoughts on that.

In these last 10 minutes, please do feel free. Thank you all for coming to the webinar. Feel free to dive in with more questions. If something comes up for you and hopefully we’ve answered a lot of your questions here, but. There is a lot of just speculation hand wringing and other things about AI and I recently went on a trip and one of my buddies like you guys are the problem.

You’re gonna end the world with this thing. I just wanted to get your thoughts, how much of how much that apprehension and panic Do you feel like is warranted. How much, how real is it that like, we just built Skynet, right? Some people have had those kinds of reactions.

What do you think about the future? What’s the next 10 years look like for AI? And should we be worried? Should we be doing something? Or should we just keep building stuff and see what happens?

Mitko Simeonov: Yeah people are going to keep doing stuff and see what happens. Some people are going to do that regardless, right?

With every technology, I think there’s like some dangers that are going to come up and AI is no exception. For example, social media, great for

Posting photos, but then a lot of misinformation and dopamine addiction to happen NSA, the internet allowed NSA to spy on all of us more effectively. Crypto allowed a lot of Ponzi schemes. A lot of biological research, can cause viruses to escape or whatever. Pandemics can happen. We just increases the existential risks, right?

And with AI, one of the existential risks that’s been touted that was Artificial General Intelligence, and we’ve seen Terminator and Skynet movies, and The Matrix some of my favorite movies that’s a really big team, and Yeah, people are afraid of artificial intelligence and talking about, is it going to be aligned with our interests?

And I think one of the biggest dangers that I see is actually, even way before we reach the point where AI is like self sufficient and autonomous and is able to create Skynet or anything like that, I think people plus AI Is always like more effective of either of those alone both for good purposes, which is what we’re trying to give a good example of pioneer, but also for nefarious purposes.

And so actually, I’ll be much more worried about how bad actors would use the AI and those and increasing this information powers much more than, for example, Skynet narrative and, communication challenges. I think there are opportunities for a lot of those models to actually.

Hackers on emotional level and subconscious level, which we don’t talk about a lot about, but I think those would be used by people that are also very skilled in understanding those dynamics. That is what I worry is going to happen much more before full on Skynet. I think reality is always much more interesting than the movies but I think just us, all of us being careful the way I would say.

The best way for us to protect against it is to actually educate most of us about it and know about the dangers. Just like we saw with social media and misinformation, the more people are aware that this thing happens, they know to like check sources. And with deepfakes that are in a lot more things are going to be deepfaked with AI and already seeing that with some of the applicants for the jobs that you’re posting I’ve seen several of them post like a very similar kind of cover letter and I could see that it’s a system and it doesn’t disqualify people, but it just becomes an arms race between various communication strategies.

I think the best we can do is just get familiar. All of us start playing with the tools and understand the capabilities of understand what they’re good at, what they’re not yet good at. And just stay current ish.

Sedale Turbovsky: Yeah, I love that. I do think, it is probably a long time before we get to Skynet, but certainly, we can see it already in scams and things that people are running where you have these, interesting capabilities to spoof people’s anything and do a lot of do a lot of potentially nefarious and damaging things.

Yeah. Yeah. Yeah. I think Go ahead. Go ahead.

Mitko Simeonov: Yeah. I wanted to just add a little bit of a positive note there as well, because it’s, that’s the perceived dangers and we as species are always more biased towards like fear and excitement. But I think there’s like a lot of great stuff to look at. For example as an innovation platform, I see large language models as being even bigger than the iPhone was with creating apps and stuff.

So I think that’s actually gonna infuse in a lot of the systems and products that we use and it’s going to make life a lot better in various different ways. And also, I expect 10 years ago, like I said neural networks were considered a failed approach. Who knows, in 10 years, we’re going to have several iterations after the current large language models.

And I think that’s going to be interesting to see what the new paradigms We don’t really know because if we knew we would just invent them right away. But future will show and then yeah, I think that’s going to enhance human creativity and especially in this domain. I’ll, I’m just like excited to see what kind of creativity we can unlock.

Sedale Turbovsky: Yeah, no I’m super excited. I alluded to this earlier, one of the challenges that we’ve seen frequently is that folks just don’t have The bandwidth to write these applications, put them in, get them through the process. And they often don’t have the resources to get support to do that either.

And I think one of the really exciting things for me about this space is that, these are the kind of tools that can really help. Streamline that and remove those barriers for people. And I, I’ve been a big fan of no code for as that movement has pushed along and watching just this take that to a whole new level where people can, build apps and learn things and just move ahead with their ideas and unlocking human potential in a really more of a democratic way is super exciting.

Yeah, I think there’s a lot of things to obviously be wary of, but there’s a lot of stuff to be excited about as well. And I think that’s really fun. I’m, we’re closing in on the end of our time here. So I just want to say thank you to everyone who came out. Hopefully we did answer your questions.

If you have more, please do send them our way. Once again, I think there’s a lot of cool opportunities here to unpack for folks who are looking for grant funding or looking to streamline some of their workflows. I will have Miko send around some links. We’ll send out an email.

This recording will also be posted up on YouTube so you can get at it later. And yeah, just thank you all. I’m going to turn it over to Mikko for any kind of like last calls to action parting thoughts that you might have, but thank you so much for joining us and for sharing your expertise.

Really appreciate it. And yeah, just. Take it away.

Mitko Simeonov: Thank you for hosting today. Thank you everyone for your time today as well, or if you’re watching this later down the road. Thank you. Thank you for watching it as well. So I have just two call to action. One is just the generic let’s all help address climate change together.

And especially in the domain of funding, the government can increase the money printer and can go burn much faster, but Making those resources being utilized effectively depends on how much all of us are putting into it. And so just getting, rolling your sleeves and trying new tools and doing something in that domain, just being the man, the proverbial man or woman in the arena.

I always want to support people like that, and I want to invite you come and join. And solving one of the biggest challenges or days. And then the other thing about Pioneer specifically, we are actively hiring for operations and engineering. So if you want to see a pioneer climate.

com, we have also a wait list. If any of you would want to use our services, but pioneer climate. com slash careers is where you can see our job opportunities. So thank you.

Sedale Turbovsky: Awesome. Thank you all. Really appreciate you all being here today. And definitely you can head over to, to OpenGrants as well.

If you pop onto Eventbrite and just follow us we do have an awesome session coming up next month with the Environmental Protection Network. Super cool. They’ve helped a variety of startups on our platform secure EPA SBIR. They’re fully funded by the federal government. So they’re great people to know because they can offer some incredible talent.

No cost services to you. And yeah, super excited to have you all here. Thank you so much for joining us. I’ll let you all get back to your day and we’ll see you next time. Thanks.